« Previous | Main | Next »

Listening to the Proms in Extra High Quality

Post categories:

Rupert Brun Rupert Brun | 16:55 UK time, Friday, 3 September 2010

Illustration for PromsXHQ, BBC Radio 3's high quality audio experiment

Editor's note: for the final week of The BBC Proms you'll be able to listen to the live radio broadcasts from The Royal Albert Hall in Extra High Quality audio. The man who came up with the idea explains the thinking behind the experiment- SB

You could make a strong case to say that of all the digital platforms used by BBC Radio, the internet delivers the best quality. We have invested over the last two years in a completely new system to code our audio for the internet using respectable bit rates and AAC coding technology. The system is known as 'Coyopa', named after one of the two Mayan gods responsible for thunderstorms. The god Yaluk makes all the lightning flashes, Coyopa makes the impressive noises.

Using our Coyopa system we distribute within the UK most of our national radio stations at 128Kb/s whilst Radio 3 is available at 192Kb/s. The audio feed for the internet distribution is derived from the signal path used to feed digital television and in the case of Radio 3 this means (like DAB) that there is no multi-band transmission processing involved. I wondered how good we could make Radio 3 sound on the internet and what people would think of it. Would it be a welcome improvement? Or given that most people don't have high quality amplifiers and speakers connected to their computers, would they even notice the difference?

The bust of Henry Wood - founder of the BBC Proms - in the lobby of the building named after him in Central London

The choice of content for the experiment was obvious to me - we had to use the BBC Proms concerts. They always attract a strong audience and this year we have a brand new outside broadcast vehicle at the Royal Albert Hall and a new link from there back to Broadcasting House. The link from the truck to Broadcasting House is linear (uncompressed) audio with a frequency response extending to 22KHz, making it one of the best audio sources available to us. The signal passes through our internal digital routing systems to the Radio 3 Continuity Suite, from where it feeds our Coyopa internet coder. After the Last Night of the Proms on 11th September the experiment will end, so listen to the stream whilst the Proms are being broadcast and please do complete the feedback form to tell us what you think of the experiment.

By happy coincidence, the experiment has been dreamt up and realised by a team of BBC staff who work in an office building called 'Henry Wood House', named after the founder of the Proms Concerts. Henry Wood House was built on the site of the Queen's Hall, which was the home of the Proms until it was destroyed by an incendiary bomb in 1941. The team walk past a bust of Sir Henry Wood as they come to work each day.

Rupert Brun is Head of Technology for BBC Audio and Music

  • Listen to Radio 3's Extra High Quality Proms audio on the Radio 3 web site during live broadcasts of The Proms until 11 September 2010. On the same page you'll find a link to a survey about the experiment. Please take a minute to complete it once you've tried the Extra High Quality experience.
  • Help us spread the word about Extra Quality Audio for the Proms by tweeting about the experiment using the hashtag #PromsXHQ.
  • Read Rupert's FAQ for answers to the big questions about PromsXHQ.
  • Read this blog post by Radio 3 Interactive Editor Gabriel Gilson on the Radio 3 blog for some additional context.
  • The picture shows audio engineer Brian Hodgson tuning audio generators at the BBC Radiophonic workshop in the 1969. It's from the BBC's picture library. Rupert worked with Brian at the Workshop.

Comments

Page 1 of 2

  • Comment number 1.

    http://www.bbc.co.uk/iplayer/episode/p009k2dc/Tony_Blackburn_29_08_2010/

    Annoys me this news immensely when you can only provide 80k mp3 that sounds shocking on my amp

  • Comment number 2.

    O.M.G. - O!. M! G!.

    I've just been to Albert Hall - and didn't leave my dinner table.

    The Mahler was just wonderful BUT - I was there!

    XHQ - RULES........

  • Comment number 3.

    Excellent.....

  • Comment number 4.

    I was listening on headphones here in the USA on a wonderful sunny afternoon in my garden. Towards the end of the Mahler 4th movement, it sounded as if one of the player's instruments malfunctioned and made a loud pop, a string or fiddle bridge I think. The realism of the noise was very surprising. The real music was even better.

    I hope this COYOPA experiment is successful and that you keep it.
    I love the BBC 3 website and all that it offers in the way of classical music.

    The USA is a dessert by comparison.

  • Comment number 5.

    Just listened to the Berlin Phil at the proms. Fantastic performance by the orchestra and what a fantastic performance by XHQ. None of the broken glass effect of DAB and none of the compression we hear on FM. I was waiting for the limiters to come on during the fff passages, but no, just wonderful dynamic range. I hope the dear old Beeb realise that they now have no option but to make the service permanent

  • Comment number 6.

    Just recorded and listened to the Mahler tonight (I'm based in London). An absolute delight. Let's not fool ourselves, the coding towards the consumer is one thing, the quality of the whole b/cast chain is another. Hence, congratulations to BBC and Radio 3! Natural dynamic sound of warmth and clarity, agree with jaybicks. No compression in the lower range, maybe slight limiter in some peaks (waveform looks a bit "cut"), but acoustically fine.

    BBC, PLEASE LET US HAVE ALL RADIO 3 PROGRAMS in this 320k format, they deserve it!

  • Comment number 7.

    Radio 3 - 320 kbps AAC - WONDERFUL! 192 isn't bad, very listenable, but at 320 all the power and the glory of the orchestra is there. Would it really be so difficult to carry on at 320 after the Proms finish? Even just for Orchestral recordings from Edinburgh, or BBC SO from the Barbican etc.? One important point - if you can do 320 during the autumn/winter concert season you should flag it up well, not just on the website but in Radio Times too. Go ahead, do it and be proud of it! There are very few Euro Classical stations at 320 mp3, let alone mp4 - (the Berlin Phil Digital Concert Hall is one) so the BBC would once again be in the technical vanguard. I'm not alone in having abandoned FM a few years ago because of the dynamic compression. What joy it has been to return with the iPlayer stream. At 320 you're bound to draw more and more listeners in, just do it!
    Jayne Lee Wilson,Liverpool UK

  • Comment number 8.

    This is a wonderful development and one I have been waiting for. I'm not sure people mainly listen through laptops - a while back I ditched DAB and moved exclusively to streaming for R3. Fed through my hifi this stream sounds absolutely superb. Given that we get HD content for TV, it seems odd that we're limited to 192 for radio (and bizarrely lower 128 for listen again). Please role 320 out permanently and also for listen again on the iplayer - an excellent way to spend the licence fee!

  • Comment number 9.

    Will you be doing a similar experiment with video - like doubling the frame rate?

  • Comment number 10.

    Briefly: the music sounds wonderfully clear and natural. I use a laptop linked up with a fairly simple (amplifier) sound system. Please continue this experiment after 9/11. Don't let the date bother you. Thank you.

  • Comment number 11.

    Is there an alternative (on Linux) to using the adobe software which is notorious for both the holes it leaves in any security (especially in any any Windoze environment) as well as allowing all the privacy violating tracking software exploited by the less scupulous companies (eg flash cookies).

  • Comment number 12.

    At last ! Real dynamic range. Whilst bemoaning the dynamic limitations of the medium,I had pretty much accepted that FM was probably as good as it was going to get, and not for much longer either.
    However, those timpani at the close of the first movement of the Mahler really made me sit up. The whole symphony was a joy, and the RAH acoustic was palpable. I only intended to sample the stream for a few minutes but I couldn't tear myself away.
    Please, please find a way to give us more of this quality on Radio 3.

  • Comment number 13.

    Magnificent! Saw the BPO last night at the Albert Hall, and tonight I get to hear them almost as clearly at home.

    I've always been impressed by the quality of the live sound the BBC manages to get at the Proms - they obviously know something about recording that most CD producers don't. But digital compression has been irritating - more so on Listen Again than live I think.

    Your new XHQ sound is a significant improvement. Thank You.

  • Comment number 14.

    Success and welcome. Hope you can keep the 320kbs feed going full-time.

    Great to hear in finer detail: particularly noticable on the quieter passages. It seemed in the Bruckner there was a more consistent balance overall with less change in audio perspective between quiet and loud passages.

    The reverb from the hall seemd very lifelike though did not come "out of the speakers" as much as I would like - the reverb actually in the hall is very characteristic of the Albert Hall. On the radio on Saturday the reverb stayed narrow.

    Having attended a few Prom concerts this year, the experience from Radio 3 is very much "the best seat in the house" even to the extent that it is impossibly clear and wide sound that is not available anywhere, front arena, stalls, Grand Tier etc.

    What one still misses on the radio is the thrill of the performance and the tension between the orchestra, arena and hall.

    The louder one plays the music the more annoying it is that the announcer and interval material is transmitted almost as loud as the full orchestra.

    Part of my enjoyment of listening to live concerts is the ability to record and relisten.
    This is gettng better on with iPlayer but the 7 day restriction is a pain.
    The ability to download and keep would be welcome

  • Comment number 15.

    Wow this anouncement is a big suprise to me. Under Mark Thompson we have seen decline in technical quality both of radio and tv. For years now the BBC has spent millions on promoting the absolutly awfull DAB radio system with it's poor sound quality and it's susceptibility to interference. The net maybe the future for radio as even some car radios have a 3g reciever.

    But why only radio 3. Please can the BBC extend this to other channels some of us would even appreciate radio 4 at 320k. What about some surround sound transmissions as well.

  • Comment number 16.

    Concur with the other supportive comments, it's a great sound. I have filled in the survey and requested more of the same please. Why not make it standard for online Radio 2, 3 and 4? That would be very much appreciated.

  • Comment number 17.

    Listening to this afternoon's "1910 Last Night" Prom what strikes most now is the smoothness of the sound; there's a lot more "air" around the upper strings and brass on the XHQ 320kbs feed compared with switching back to DAB or Radio 3 on Freeview (with matched levels). Those both sound rough. Particulary apparent on Dvořák's Humoresque.

    Very welcome too are the clean sibilants of the female vocalist and the presenter, not slurring as on all other feeds available.

    Lots of opportunties to hear hall reverb for the Lenonora #3 and it's sounding great withut muddying the direct sound.

    Lots of peak compression on the climaxes of the Wagner Kaisermarsch: did the balance engineer set the levels a bit hot and get caught out at the beginning? (Whoever you are today, thanks to you and all your colleagues and assistants, you do a great job).

    On my equipment, the 320kbps feed is significantly later still compared to both Freeview Radio 3 and DAB. And of course far later than FM, which today is cursed with warbling birdies on my equipment here in West London.

    I'll be listening to the second part on the XHQ 320kbs feed only and forgetting the technicalities. Need I say more?

  • Comment number 18.

    I agree with many of the statements concerning the improvements demonstrated by XHD. I am listening on high quality open headphones and concurr - a huge improvement; I am quite used to the acoustic and ambient noise of the RAH as I have spent many hours in there recording the annual School's Proms and this is amply shown up with XHD! Please do keep the service and extend it to cover R3 and R4 throughout the year! I have been dissatisfied with the thinly spread bandwidth available to a DAB multiplex though the potential might have been good with less channels. Keep up the good work - some of your audience really do appreciate it. Mike McMillan

  • Comment number 19.

    To echo other comments, please can we have more of this.
    I have been truly disappointed with the quality of DAB. Frequency response and dynamic range may measure well, but it just sounds wrong. I thought I was just being too sensitive to things until my daughter complained that Radio 5 Live is less acceptable on DAB than AM, despite all the crackles and whistles on there.
    I have been using the 192k internet feed as an alternative to FM, but it lacks some clarity that is there in the FM feed despite some unpleasant multi-path effects that annoy me.
    The time has come for a change of plan. Instead of an all-too-soon analogue turn off, please scrap the current DAB system and replace it with something that provides quality sound.
    This experiment is a major success. The sound is spectacularly better than anything currently available. Can anyone explain why such a moderate increase in bit rate can make even speech sound better?
    Once again, please more of this. It allows me to listen to the music instead of listening to the radio.
    Eric

  • Comment number 20.

    Eric_T commented thus:
    Can anyone explain why such a moderate increase in bit rate can make even speech sound better?
    I can only add that though the intelligence in speech may be carried over a narrow bandwith system (such as a telephone line), the natural sound one expects to hear does require far more bandwidth to capture and reproduce. I would go as far as to say that speech requires the full bandwidth like music. I have carried out experiments on data compression for speech and music, speech appears to suffer more than music on some systems!
    Mike McMillan

  • Comment number 21.

    I am delighted with the idea of High Quality sound via the Internet, and I am glad that the BBC is catching up with some of the Dutch and German stations that already offer this.

    There is just one thing that I need and that is a URL than I can enter into vtuner so that my various internet radios - particularly my HiFi system can make use of the service.

    When will the BBC publish a URL for the High Quality stream that an Internet radio can use?

  • Comment number 22.

    Sorry to dampen the enthusiasm, but why is the audio signal passed through a dynamic limiter before being made available as 'extra high quality'? Regarding dynamics, DAB retains the upper hand at present. Do please stop truncating the peaks.

  • Comment number 23.

    Just listened to the Ulster Orchestra Prom via Proms XHQ through my hi-fi and the sound was fabulous. The most noticeable improvement was in the bass sounds, but overall it sounded far cleaner and superior to DAB, Sky or Freeview, and comparable to FM without hissing or interference. The Last Night festivities should sound spectacular.

    This without doubt leapfrogs DAB and is radio broadcasting for the 21st century - true high definition radio. I hope it becomes the international standard for radio broadcasting. At the very least, it should be back for next year's Proms.

  • Comment number 24.

    I listened to the Mahler, and to the Sibelius and Parry, and I'm sold. This beats anything else on offer, as far as broadcast, downloads and CDs. And I'm listening on a simple £300 radio player with integral speakers. So please make it permanent! And if it's held up by a financial limitation, then cut Chris Evans' hours in half. Or fire all the football pundits. This must go on!

  • Comment number 25.

    Friday evening.

    It is very early days so these are only my initial observations and comments. On Friday evening I just had time to extend the decoded bit-stream to the HiFi equipment in the front room and listen on the B&W 801 speakers to the Mahler 1. My only comparison on Friday was FM from Winter Hill (4 miles away). Maybe the occasion was partly to do with it (BPO/Rattle/Mahler 1) but the 320 kbps stream sounded magnificent - the detail, clarity and precision of attack was wonderful. Of course the instruments could heard with great detail and clarity but also the sense of three-dimensional space around them too - wonderful transparency.

    During a quieter section half way through the symphony, I switched over to FM (perfect reception here) and carefully matched audio levels. I was quite astonished at how much dynamic range compression was being applied to the analogue FM feed. It sounded quite squashed and unexciting by comparison. Until recently I had always regarded FM as less fatiguing to listen to and having a clearer sound than any other BBC radio source (including the 192 kbps mp2 feeds on digital satellite TV, digital terrestrial TV and DAB - especially those utilising a 128 kbps bit-rate - UGH!).

    More recently I have come to regard the 192 kbps AAC feed of Radio 3 as superior to FM and the highest quality audio of any BBC radio output. There are still problems with these streams but it is not the the fault of the encoder. Frequently the audio level to the encoder is set too high together with hard limiting of the output to -6 dB of peak level - all the peaks are thus rounded off. This situation is worse on the other 128 kbps AAC network streams where nasty distortion akin to clipping occurs on such as Radio 4 live speech. (The digital TV radio feeds are fed at a more appropriate level and don't suffer from this).

    Also why is it the 'Listen again' feeds are more heavily processed and limited prior to reaching the encoder? They are so much worse than the 'Listen now' live feeds. I wonder how the repeats of these last few Proms will fare in this regard.

    I recorded the decoded 320 kbps stream as a wave file. Looking at it I would still say the audio fed into the encoder was a few dB (6 dB?) too high. During the loud sections everything was hard limited to - 3.6 dB and flattened off (of course FM was much worse in this regard).

    I have yet to make the most relevant comparison. I intend to record both the 192 and 320 kbps AAC streams simultaneously for later comparison but I can only do one at at time on my main PC. Tomorrow I will set up a second PC to download and record the other stream provided the evening broadband contention ratio doesn't foul up the transfer of 192 and 320 kbps at the same time. I normally get a 6.5 Mbps download speed but that hasn't provided a guaranteed solid feed of the 192 kbps stream without the occasional break and re-buffering. I will then to be in a position to do more carefully controlled listening tests to evaluate the hoped for benefits of the higher bit-rate.

    Further feedback pending ....

  • Comment number 26.

    Sunday evening update – there IS a problem with levels. Please fix.

    Well, I tried setting up two alternative PCs to record online streams but it wasn’t as straightforward as I had hoped. The on-board audio only gave the option of an already decoded D/A stereo-mix and the audio quality was significantly inferior to that on my main PC. I am going to have to acquire a dedicated PCI card to make valid parallel recordings of the two streams.

    However on Sunday evening I recorded a reference mp2 file from digital satellite of parts of the concert, subsequently converted to a wav file. I also recorded the online streams in real-time as wav files, necessarily one at a time in chunks of 5 or 10 minutes, while listening to them. What was immediately apparent was the much higher audio level of the 320 kbps versus the 192 kbps stream. It is well known when switching between audio feeds the one which is a little higher in level, other things being the same, sounds subjectively better. But here the level discrepancy was such as to have me leaping for the volume control.

    I have established is the 320 kbps stream is 6.9 dB higher in level that either the 192 kbps stream or the digital satellite mp2 feed. Tonight the latter two appear to be within 0.1 dB of each other. The 320 kbps stream is also hard limited to about – 3.6 dB of peak level.

    In the Rachmaninov Piano Concerto on the digital satellite recording the highest level peaks at -5.2 dB of maximum level but the 320 kbps stream recording at the identical point is only – 3.5 dB. Given that the level of the later should be 6.9 dB higher this means the top 5.2 dB of the audio dynamic range is being clipped (OK rounded) off. There is no evidence of peak limiting on either the digital satellite feed or the 192 kbps stream. (Limiters must surely exist to protect against system clipping but the levels fed are appropriate and I have seen no instances of them being invoked).

    Who is responsible for maintaining levels? For this PromsXHQ experiment to have validity it is incumbent on those running it not to have the high level audio bashing the limiters all the time. For the most part their subjective affect is quite benign but, if levels were correctly set, the impairment thus caused would not be a factor in our evaluation. The whole rationale of this experiment is to provide the highest audio quality – subjectively indistinguishable from what leaves the mixing desk at the RAH.

    THIS NEEDS TO BE ATTENDED TO - before we are much further into the trial.

    Other than the above observation I would say the online streams are both significantly better than digital satellite (and, from previous observations, also better than the Freeview radio version). The dynamic range compression I observed on FM on Friday was much more severe than I previously remember and I would have to considerably downgrade its perceived audio quality. The audio quality of the 320 kbps stream, other than as noted above, was a revelation. It has left me needing to find some new vocabulary to adequately express what I have heard.

    More feedback when I have had the opportunity to directly compare the two online streams ….

  • Comment number 27.

    To add a human perspective, I listened to Prom 66 (Berlin Phil, Wagner etc) on EHQ, the first half on Bose Companion 3 speakers and the second on Sennheiser PXC 450 headphones, and found it more involving than the normal streams. It reminded me of the difference between watching a Prom on HD and standard TV.

    Currently (1025 BST 6 Sep) listening to Sarah Walker Classical Collection. The EHQ stream is perfect, whilst the normal R3 stream has a horrible burbling and lots of dropouts. Not sure how EHQ the content is but it sounds pretty good. The stereo is brilliant.

    PS As of 1105 BST, the standard stream is still burbling but for the brief time it stopped, I was able to make an AB comparison and the difference between EHQ and standard is striking.

  • Comment number 28.

    With regard to message four above - it was a harp string which broke.

    David

  • Comment number 29.

    Thanks everyone for the feedback, which is extremely useful. The main concern seems to be clipping of high level transients. Like so many things, getting this right is a matter of compromise. If we have too much headroom we effectively waste a few bits and throw away some dynamic range; if we have too little headroom the clipping becomes unpleasant. We also have to consider what happens to the audio further downstream - every process we apply to a digital signal has the potential to create overshoots which can become clipped further along the chain.

    As well as trying to use the right amount of headroom, we also use good quality audio limiters as protection. For the BBC Proms, there is a limiter on the mixing desk at the Royal Albert Hall, to catch any transients that would clip on the circuit to Broadcasting House. There's another limiter in the mixing desk in the Radio 3 transmission studio (continuity suite) to deal with any peaks here. Both limiters are set to provide protection for high level peaks only, not to modify the dynamic range. The signal is then fed into our Coyopa system for coding at 320kb/s without any further limiting. None of the digital platforms have multi-band transmission processing (compression) on Radio 3.

    Our normal internet distribution at 192kb/s has been well received; the consensus seems to be that the new feed is even better which is very encouraging. The fact that some people are noticing a little peak limiting may mean that we need slightly more headroom at some point in the signal path and we will look at this. Please keep listening and giving feedback; this is a real experiment and I didn’t' expect to get it perfect first time!

    Rupert Brun, Head of Technology. BBC Audio & Music.

  • Comment number 30.

    personally I think Radio 3 should be 80k MP3,the same quality you think is good enough for BBC Local listen again.

    320k aac should be used across the genre of music styles,not just classical. Rant over!!!!

  • Comment number 31.

    Rupert, thank you for your reaction to our feedback; it is indeed satisfying to know that someone is listening (or reading!) to the customers or listeners. I am a sound engineer rather than a computer engineer and though I have a reasonable knowledge of data streaming etc, I am by no means certain I won't put my foot in it, but here goes:
    In a previous posting, I alluded to my dissatisfaction with DAB for although the architechture would allow good audio quality - it is at the expense of bandwidth; sharing the available bandwidth between 3 or 4 channels is fine but with as many as are available on a multiplex, there is much reducing of bandwidth per channel and some in mono only to save a little extra again.
    Using greater bandwidth (thus greater data transfer rates) as you are doing for the experimental XHQ 'Last week of the Proms' suggests to me that you are imposing an increased load on the servers and the infrastructure 'tween you and the listener. If for arguments sake this greater bandwidth were to be used all the time and extended to other networks (R4 for a start), would we then find a dramatic increase in hardware and network speeds is required? What impact might this have (I'm thinking of servers etc.) on the carbon footprint?
    I think I am right in saying that the bandwidth available to you for DAB is restricted by outside factors and though it may be technically possible to provide 320kb/s on 8 or more networks, you are not permitted to use such a large chunk of the available resource for DAB. Will such a restriction disappear when propogation takes place over the internet?
    Regards and Thanks, Mike McMillan

  • Comment number 32.

    As a long-standing listener to BBC radio -- and occasional critic -- I too have been favourably impressed by the overall sound using 320kb/s AAC: it presented a clearly better spatial quality, but also has some awkward moments of peak limiting. I use a high-quality separate D/A converter to listen to digital sources, through Quad electrostatic loudspeakers.

    The choice between now much compressed FM and very disappointing DAB sound has reduced my listening to Radio 3 in recent years almost to zero. The rise of online availability has to some extent restored my confidence, despite the often irritating presenters whose voices are additionally always far too loud.

    I agree with the comments about speech reproduction. I well remember the early use by the BBC of digital recording of concerts in the late 1970s, when I was surprised by the unexpected improvement in reproduction of the bass end of music and also of the announcers' speech, which at last sounded as good as in live broadcasts. We are (naturally) very sensitive to tiny errors in speech sound; the old pre-digital long-distance analogue distribution lines with their delayed treble frequently produced an effect like loose dentures (as my mother commented).

    One perennial problem for the BBC over the decades has been the battle between sound quality, available bandwidth and sheer intelligibility. This has resulted in the ruination of (successively) basically good AM and FM technologies and also DAB (which I must say started off fairly badly in the first place). AM and FM are now, notoriously, badly compromised in bandwidth and/or dynamics: DAB in its present form varies in sound between simply boring (Radio 3) and downright disgusting (nearly all the others). As has been commented, the ordinary AAC streams sometimes have an inconsistent hard clipping well below peak level, which causes distortion on speech. I hope that this experiment in online high-quality access may be allowed to develop, despite the inherent problem of internet connection quality and continuity (my listening has been interrupted a number of times by network congestion).

    Lastly, I should comment that the realism of Radio 3 sound hereby appears at least comparable with the best of 1970s FM -- I have several nice old off-air tapes (er, cassettes -- with Dolby B :)

  • Comment number 33.

    Thanks indeed for the informative feedback re. the likely source of the limiting some of us have noted, Rupert. Certainly the tonal quality and the clarity of transients is far superior in the 320kbps aac-lc offering to that from DAB. DAB, however, does seem to offer more clearly defined peak dynamics. I wonder if the relative levelling of peak dynamics is something inherent in the Coyopa system?

    I will have a go at using a DAW to make dynamic 'fingerprints' of the same material in 320kbps aac-lc and 192kbps DAB mp2 versions of the same material and post links to them here later, if that's o.k.?

  • Comment number 34.

    Great feed. Currently listening to Jazz on 3. The clean sound here and on the late night Prom immediately before is a real treat to the ears (especially the Prom). The sad effect of the split band processing on the FM feed is all too clear on direct comparison.

    Let's hope that the BBC decides to keep the XHQ feed for Radio 3, and fine-tunes the level issues.

    (But isn't it sad that one can't listen to a studio quality feed on Radio 3 FM?)

    And many grateful thanks to those who made the experiment possible.

  • Comment number 35.

    Thank you very much Rupert for your feedback back to us. It is greatly appreciated.

    While I agree it is very important to optimise the levels to take advantage of the dynamic range available and, at the same time, not incur any signal clipping, I don’t think this is the problem here. All the digital transmission paths available to me (digital satellite, terrestrial Freeview radio, the 192 kbps/320 kbps aac streams and also the infamous DAB) utilise 16-bit resolution with 48 kHz sampling. They must all have hard limiters to avoid full scale peak level clipping. The problem simply appears to be the level being fed on the 320 kbps stream is 6.9 dB too high.

    This evening, on my main PC, I recorded Radio 3 from both digital satellite and terrestrial Freeview as mp2 files simultaneously while, at the same time the decoded 320 kbps aac stream as a wav file. The mp2 files were later transcoded to wav files. On another PC I recorded the 192 kbps aac stream also as a wav file.

    The later was transferred to the DAW on the main PC and all the waveforms examined. This revealed the levels on all transmission paths matched to within 0.1 dB with the exception of the 320 kbps aac stream (I didn’t look at DAB or record FM). The 6.9 dB higher level here would have caused system clipping had it not been for the intervention of the hard limiters.

    I can only speculate and surmise but presumably the linear uncompressed link (probably 24-bit resolution?) between the RAH and BH is common to all the digital feeds and streams. It cannot then be the limiters on the output of the mixing desk of the OB vehicle at the RAH. Currently, although the experiment is called PromsXHQ, the high bit-rate stream appears to be running 24/7. So the output of Radio 3 Continuity is being continuously fed to the Coyopa coder but somehow the level is being raised by 6.9 dB and this higher level audio content is continuously hitting Continuity’s limiters? It is almost indicative of an un-terminated signal path somewhere in the continuity desk.

    Notwithstanding the technicalities, this evening demonstrated another wonderful listening experience. The Beethoven Piano Concerto 5 was a delight and, for the most part, I forgot about the limiting problem as so much else was benefiting form the greater openness and clarity of the sound. We must surely be hearing sound indistinguishable from that at the output of the RAH mixing desk – all it needs is the level/limiting problem to be attended to.

    I wish to heartily thank Rupert Brun and his team for making this possible. In the past the BBC stood for technical excellence but of late this image has become severely tarnished. The DAB fiasco is a case in point. It is a great pity radio spectrum has been sold off so it will never, in the future, be possible to broadcast such high quality audio and receive it on a radio via an aerial. Internet radio is definitely in the ascendancy now.

    Someone else has commented the degradation to FM by dynamic range compression and the inferior sound of the UK mp2 broadcasts had reduced his listening to Radio 3. I too have spent a lot more time listening to the 320 kbps mp2 German classical satellite stations – after all language isn’t much of a barrier for classical music – but the introduction of high quality streams on a permanent basis would change all that. Here’s hoping.

  • Comment number 36.

    It’s great that this experiment is continuing and I hope will become permanent once the teething troubles are shaken down.

    I already use the different feeds for different purposes and this increases my listening to BBC Radio overall:

    FM/VHF is great in a car, where I am glad to have the benefit of a high-quality and sensitively-managed compression chain rather than something provided by a car manufacturer that has to suit every audio user of every taste. I won’t be throwing away my VHF/FM receivers just yet precisely because it offers a sensitively compressed version of the programme which is useful for casual listening in a car and around the house.

    DAB has an option for dynamic range encoding at source but my experience is that this is relatively primitive.

    How wonderful that we now have available at home something even closer to the sound as heard by the balance engineer, without the physical constraints of working on location or the distractions and responsibilities of that job.

    Let’s remember these different audiences and their needs, treasure the extra high quality feed and enjoy it for concentrated home listening but not abuse our privileged position in the BBC’s data budget. The listener who is commenting on the low data rate of local radio has a point. And let’s not forget that the BBC is currently fighting to balance its financial budget and that DAB initially launched with Radio 3 at 320 kbps albeit with an mp2 codec which the more sophisticated AAC codecs have been designed to improve upon. That privileged position was eroded due to politics. I anticipate we’d find that 320kpbs mps would not sound as good as 320kbps AAC.

    There used to be a method of working where the balance engineer listened off-air and balanced including the entire transmission chain, out to the transmitter and back. That made the fullest use of the transmission channel. That way of working stopped some years ago but was commonplace when the NICAM distribution chain was first introduced so that the limiters doing most of the work were the NICAM input limiters and everything prior to them in the chain was analogue. Those days are gone now and that way of working is probably infeasible now due to the codec delays. But some of our treasured home recordings will have benefited from this technique.

    What’s my point, get it right for Radio 3 and then hope that the BBC can find ways for this experience to benefit all audiences

    There remains the problem of how to connect a computer to the hi-fi equipment. Clearly some of us have arranged this successfully already. I have been horrified at some of the arrangements used by my less-technical friends and this has come to light particularly after tipping them off about this experimental feed over this past weekend.

    Problems of earth loops (there’s a real safety issue here with inexperienced people lifting earths to try to reduce buzz), fan noise, induction noise from screens and of course hard disc noise.

    There’s a gap waiting to be filled of a realistically-priced receiver unit to deliver really good audio from the internet to domestic audio equipment.

    Meanwhile, some guidance on air would be welcome: maybe an item for “CD Review”?

  • Comment number 37.

    As @mgalley has pointed out there's a level problem at times with the experimental feed. I want to apologise for this - it's an experiment, I knew we would learn a lot but this is a problem I wasn't expecting. We carefully set the levels to match and when we check again later, there's sometimes a difference of 6dB. The engineers are working on this but pinning down exactly what is causing it is taking a little while. Our main transmission chains are duplicated, as is the Coyopa system, in order that we can undertake maintenance and upgrades without falling off the air. There are a large number of interconnections between the various items of equipment and so the whole system is quite complex to trouble-shoot when intermittent problems arise.

    Rupert Brun, Head of Technology, BBC Audio & Music.

  • Comment number 38.

    May I just add my thanks to Rupert and the team for allowing us to take part in this experiment? I use a laptop to feed a line level signal to B&W 805s via a pair of Arcam amplifiers and I consider the resultant sound to be at least as good as from my best CDs. If this feed were to become a "production" offering, I'd spend money on a proper, silent, source.

  • Comment number 39.

    Listening to La Mer. The 320 stream is running almost 7dB louder than the 192 stream, and not surprisingly the high level transients are ruined, everything ruthlessly clipped at -4dB. On the 192 stream, there is no visible or audible sign of limiting. Please, please sort it out. This is the only thing (plus a couple of momentary dropouts) spoiling an otherwise wonderful audio experience.

    Actually there is another problem. Question: How do you make one person sound louder than a whole symphony orchestra? Answer: By making her a Radio 3 announcer. She is too damned loud, by a factor of at least 2.

  • Comment number 40.

    cantobel, by "the 192 stream", do you mean the 192kbps mp2 stream offered by DAB, Freeview and satellite? If you mean the standard iPlayer stream, I think you will find that that is now 128kbps aac, though 192kbps mp3 is used for the iPhone version of the 'Listen Again' iPlayer file.

  • Comment number 41.

    By the way, those contributing to or just reading these comments might find the discussion at:

    http://www.bbc.co.uk/dna/mbradio3/F7497566?thread=7737460&skip=0&show=20

    to be of interest.

  • Comment number 42.

    Nethersage, as far as I can see, the live AAC Radio 3 stream (which is what I was referring to) is at 192kbps AAC, just as it always is. The other national stations are at 128 AAC. The Listen Again AAC Radio 3 stream shows as 128, though I have been assured it is actually also 192. Freeview and satellite R3 are always 192kbps mp2, DAB R3 is usually 192 mp2, sometimes 160.

  • Comment number 43.

    Ah, shows how often I access the live version of the iPlayer for Radio 3. I had assumed the labelling of the 'Listen Again' version as 128kbps aac was correct, and would also apply to the 'live' offering. Clearly not, as I have just observed. Strange about the 128kbps label on the 'Listen Again' version though.

  • Comment number 44.

    We have found the cause of the 6dB level difference. Our Coyopa system has two halves, each having over sixty servers. One of these servers wasn’t running the correct version of one part of the software, as a result of which it wasn’t picking up the correct gain structure from the control system. The reason the fault appeared intermittent is that the fault was only present on one half of Coyopa – when you clicked on “play” you might have picked up the feed from the “good” system or the “faulty” one and there was no way of knowing which you had got.

    In response to the feedback received about peak clipping we have also made a small adjustment to the gain structure immediately before the codec. I hope you all enjoy tonight’s Prom and please let me know what you think of the new settings.

  • Comment number 45.

    Thank you so much for this 320kbps trial. I've been listening to it via Squeezebox feeding the digital input of my audio system and it has been a real joy to listen to. Please can we have this option permanently available?

    Thank you also for your efforts to deal with the peak clipping, etc.

    But it's worth pointing out that there is also noticeable limiting or clipping on the Radio 3 'Listen Again' streams. This seems to have crept in sometime since spring of last year. Perhaps you could refer this to one of your colleagues.

    (Apologies for going off topic.)

  • Comment number 46.

    Thanks so much to Rupert and his team for solving the problem. The Bruckner 7 tonight was just magnificent. No clipping at all. There was a 7ms dropout in the 2nd movement which, on examination, did not lose any data, just was a 7ms pause, which is now removed, leaving a perfect recording.

    Thanks also for deciding to stick with 2-channel stereo for this experiment, rather than following some continental stations (such as Bayern 4 Klassik) which have moved to surround sound for their premium service. 2 channels, 2 ears and a good pair of headphones remain the best option.

  • Comment number 47.

    Dear Rupert,
    As an audio consultant and general wide user of very high-quality hi-fi equipment over many years, this new move into true high-fidelity sound reproduction for radio is just what I have been waiting for. However, for some time now, the DAB digital radio system has been in my opinion, a very poorly thought-out solution for radio listening, where sound quality is concerned. It has been 'pushed' onto the public through much hype and publicity as being 'better' than FM, which is not the case at all. The technology behind DAB was out of date before it was even launched in the 1990's with it's inefficient packaging of data, ultimately requiring much higher capacity to carry quality sound, which initially started off with half-measures of 'fair' quality sound, only to be reduced when more stations were added to the network. It was radio three listeners, listening at home on good hi-fi equipment that were able to identify FM offered the best sound. As with any terrestrial broadcast, a good aerial is required and FM reception has never been a problem when correctly set up. However, the current test transmissions via 'internet'radio of the Henry Wood 'Promenade' concert broadcasts, (I have been listening via an outboard DAC for the best results possible), are a reminder of the sound quality once enjoyed on FM before the Optimod 'dynamic-range' limiter was used in force. As one other correspondent commented, the broadcasts of the 1970's and 80's on the VHF/FM system were regarded as offering the listener the best sound quality one could obtain on the very best hi-fi equipment; live broadcasts, (not just radio three), from the BBC were regarded as the 'cutting-edge' in true quality sound. Many people today think that only recently has good sound been available, when in fact it has been around for many decades. Today though, the means to reproduce recorded sound accurately is quite the reverse, with smaller speakers to suit modern interior tastes being used and less obvious equipment implemented,(mini systems), and now the MP3 'on-the-move' revolution with sound-docks becoming most people's idea of 'hi-fi' in their homes. For many, the 'best' hi-fi today is in the car; hence the reversal of the 'switch-off' decision of the FM system, because DAB has issues with reception on the move. So, as FM is still the most widely used source for radio, and a platform that can be used for the highest sound quality when freed of dynamic compression applied by broadcasters and proper aerials implemented by users of good hi-fi systems, a system which does not have issues with relying on broadband connection speeds, (surely an issue with 'hi'res' audio on a multiple-station network), why not offer everyone this level of quality, as per your most brilliant sounding test transmissions via the net, on the FM services, such as radio's 2,3 and 4? The optimod system was there to address listening in the car, (most cars are very quiet inside today, so coping better with ppp-fff swing in dynamic), and switching off DAB, (which we all know the reasons why it was conceived)? Internet radio is the way to go in the long run and I certainly use it for my clients, but whilst there are many like me who have very expensive FM tuners and hi-fi equipment and many who listen on portable audio and car audio, FM is still a reliable platform that serves the wider audience at large very well and should not be pushed away to suit government strategies this early on. I know that my comments may sound at odds with current thinking, but I hope you will post these all the same as merely 'my personal thoughts'. Thank you for a most interesting development in broadcasting.

  • Comment number 48.

    Thank you Rupert for your dedication in wrestling with this unusual level/limiting problem. I can confirm it is fixed. I wondered why you had thought the problem was intermittent when, at all times I listen to the XHQ stream, it was high level. This afternoon I listened to the Prom on Afternoon Performance and concluded the 192 and 320 kbps streams were subjectively at the same level.

    There was another difference. I gained the impression that little or no dynamic range compression had been applied. This is in contrast to the repeat of the Czech PO, Dvorak 8 broadcast the previous afternoon. I know it has been usual to compress the afternoon programmes as I found out in previous years when re-recording a repeated Prom. I will have to wait until Friday for the repeat of last Saturday's Prom to verify this. Maybe you could comment on this?

    I am intrigued by your 'small adjustment to the gain structure immediately before the codec'. I checked levels on Freesat/Sky, Freeview and the 192 and 320 kbps streams. I now find, as before, Freesat/Sky and Freeview are within 0.1 dB of each other. The 192 and 320 streams are also within 0.1 dB of each other but both are now 1.9 dB higher in level than the other two sources. Is this something to do with the 'adjustment to the gain structure'?

    Not withstanding the technicalities tonight's Bruckner was magnificent on the XHQ stream - wide dynamic range and with no sense of strain during the loudest climaxes. The Bruckner wall of sound was full of nuances and detail. As others have commented the naturalness of the human voice is also greatly improved. It is so easy to listen to without the mp2 'grunge' on male speech and the angular spread and sandpaper effect on sibilants.

    Yes, one complaint - when the audio quality is as superb as this it does encourage listening at higher levels. This is fine until the announcements come along and, guess what, they are much too loud!

    Apart from the last comment you and the BBC are to be congratulated for offering, if only fleetingly, what is possible by way of delivery of truly high quality audio to the home. Internet bandwidth/bit-rate considerations should surely progressively lessen (otherwise how could the BBC already offer streamed HDTV). Maybe the time has come to be more aggressively promoting internet radio especially if all BBC radio sources could be streamed at the XHQ rate.

  • Comment number 49.

    Yes this 320kpbs AAC feed is fantastic. I urge for it to become permanent.

    I concur with the decision to stay with two-channel sound as the main mix.

    Surround sound can sound great and AAC can support it whether 4.0 or 5.0 or 5.1. However there is a huge temptation to balance for surround and fold down for stereo, despite the stereo audience being by far the larger radio audience.

    This discussion becomes more complex if television is present (as at some Proms) as the HD television audience with access to surround sound might be judged larger than the radio audience. Ideally there should be two separate balances made but this may not always be financially viable.

    I’d suggest that the approach should be to consider the master sound the radio balance and add surround for the HD television audience with access to surround sound. Television already has a number of contradictions between the sound and the pictures whereas the radio listener is wholly dependent on the audio, so should have the best sound available.

    The piece from this week’s music (so far) which I’d use for demonstrations is the Debussy Prélude à L'après-midi d'un faune from Tuesday’s Orchestre National de France Prom. The electric attentiveness of the audience are outstanding. The triangle is fantastic in the new XHQ sound and the medium close perspective allows an oustanding low electronic noise floor without compromising the tone of the instruments.

    And BBC presenters are always too loud in the balance and always have been at least since the late 1970's. Why not pull back the applause far more just before the presenter speaks? Where I have made (or post-produced) a recording for repeated personal listening, a fade of about -6dB suits me fine. But this is an old problem: I’ve been doing something similar since mastering the coordination to match fades on the two level knobs of a Revox A77…

    Thanks again to all at the BBC for making this happen.

  • Comment number 50.

    Tremendously impressed with XHQ. I took the optical output from my lap-top (PowerBook G4) into an Audio Design DMM1 digital mixer and put the Astra satellite 192 kb/s Layer II in on the other input. Used the same digital to analogue converter for both signals, so this is a strict A/B comparison.

    XHQ is smoother and goes a long way to getting rid of the listening fatigue of 192 kb/s Layer II that puts me off listening to Radio 3 for long on either DAB or Satellite. A real joy. It's also very much better than FM, which is very heavily level compressed.

    Wonderful to have the BBC leading in transmission quality again.

    Of course XHQ reveals even more accurately the artificiality of many Proms balances, with a vague stereo image and excessive use of close microphones in the mix. But the transmission system itself seems good, as confirmed by the sound of the Simon Preston RAH Organ filler last night.

    Congratulations, Rupert and the team- if the experiment is continued I will definitely listen to Radio 3 more.

  • Comment number 51.

    Thanks to everyone for taking time to provide feedback. To answer the questions about the gain adjustments and why the 320 and 192 streams now match - we adjusted both to have slightly more headroom. We have made the same adjustment to the systems used to record the "listen again" files so you should find the quality improved for listen again programmes recorded after about 4pm yesterday. The reason the 320 stream was always “too high” was that the fault was with one of the coders for the 192 stream which was 6dB too low. In essence, both 320 streams and one 192 stream matched each other and although performing as intended, the consensus was that the clipping was unacceptable. The “faulty” 192 stream was, I would argue, too low and was wasting dynamic range. As a result of this experiment I have now set all the streams and the recorders for “listen again” to a new level which your feedback so far indicates is a good compromise. I think it interesting that an experiment to compare 192 and 320kb/s AAC has so far been focussed on headroom which, apart from the fault, was the same at both bit rates. Now we have the gain matched for both, at a level where (so far) clipping isn’t be a problem, I look forward to your thoughts on the difference between the bit rates.

    Rupert Brun, Head of Technology, BBC Audio & Music.

  • Comment number 52.

    Rubert,

    I have a bone to pick with you. As an inveterate nit-picker you are removing all the nits I can pick!

    Seriously though, I thought last night's Prom broadcast on the XHQ stream was remarkably good. Also the dynamic range was wonderful - as good as on many CDs.

    When the XHQ stream started I compared it with the iPlayer stream and adjusted the volume controls on the windows to make the levels roughly comparable. Listening just on a laptop with Sennheiser HD600 headphones to XHQ, the music felt much more 'real', natural and, most importantly, engaging to listen to. When I switched to the iPlayer the music sounded 'flat', somewhat lifeless, the timbres seemed somehow lacking and there was a much reduced sense of ambiance and detail.

    (Since then I have been using Squeezebox to relay the stream to the DAC in my audio system.)

    Thank you for engaging with us and taking on the points raised in such an open way. It is very much appreciated.

    Thank you also for dealing with the Listen Again clipping (and for making the XHQ accessible 24/7 during this week).

    This has been a superbly useful trial and I really do hope that the XHQ stream can be made available on a permanent basis in the near future.

  • Comment number 53.

    By the way, the XHQ stream doesn't seem to be available for this afternoon's Prom repeat. At least not from the http://www.bbc.co.uk/proms/2010/audioexperiment/

    Is this a bug or is it now a policy to limit its availability to the "live" Proms performances?

  • Comment number 54.

    Just a further point.

    When accessing the XHQ stream via Squeezebox it appears that the URL changes at 4:00pm. This means that the music abruptly stops, even though the Proms repeat hasn't necessarily finished, and to regain the music I have to re-access the stream. I'm not sure whether this also happens when listening directly to the stream on the Proms website.

    Perhaps the time for the change of URL could be delayed to 5pm (the start of In Tune).

  • Comment number 55.

    Re My message No 53.

    Apologies - this seems to be a problem related to the upgrade of Firefox that happened first thing this morning.

    The XHQ stream still plays fine in IE8.

    (The point about the URL seeming to change at 4:00pm still stands though.)

  • Comment number 56.

    Just in case you missed it Rupert has written a further post explaining some more detail which you can find on the Research and Development blog.

  • Comment number 57.

    First of all a huge thank you to Rupert and his team for setting up the XHQ experiment and for this fascinating and instructive exchange, both of which I have been enjoying for the past few days.
    Next, to respond to your request this morning Rupert, but surely it is already clear: yes the 320 aac feed is significantly superior to 192. Having never chosen to listen to radio 3 live via the web except when in my Paris flat I set up both streams today and 192 is crude and scrawny in comparison - quite aggressive at times: 320 much more detailed, airy and spacious, timbres more true - and, the whole object of the exercise, one is in much better contact with what the musicians are doing.
    Last night I decided to record the Bruckner so as to burn a CD and play it on the best system in the house. A good job I was recording it as it hung quite a few times despite my having a theoretical line speed of 6496 kbps. (For the first few days I had no such problems. Is the popularity of this experiment causing extra demand?) Such difficulties and the quality of standard sound cards will be limiting factors in many people's enjoyment of this service I fear. After 90 minutes spent excising the gaps I was rewarded with a magnificently transparent sound, superb detail, including the precisely delineated coughs, a wide dynamic range - and the performance was pretty special too!
    However, ....where it seems I differ from most contributors to this discussion is in my reactions to the 320 aac feed as compared with fm. In terms of silent background and available decibels 320 aac is faultless, but it is a lossy codec and to my ears a price is paid. On the already high quality system attached to the computer I have consistently found the sound to be detailed certainly, especially in the treble, and also with slightly superior extreme bass, but lacking in depth and with much poorer timbre than fm - this with an expensive sound card fitted in the computer, feeding into an Audiolab pre, Quad power amp and very wide range speakers. Comparing 192, 320 and fm (from a Quad FM4) on Jennifer Pike's recital this afternoon the limitations of aac as a pure 'transmission channel' were very clear - the 'richness' of timbre and the way violin and piano expanded into the acoustic on fm were a revelation in comparison with either aac feed. The same was true a few seconds ago of Renée Fleming's (speaking) voice. I recorded the Bruckner so as to hear the 320 feed at its best, where the source would be in some respects superior, on the best system in the house, with particularly good spatial imaging capability. But I have to say it confirmed what I had already heard - timbre judged second by second seemed beautiful, but somehow after a time one could become slightly indifferent. There was nothing like the magical sense of being transported to a different acoustic that properly done FM can provide. Whilst there was a degree of depth, it was meagre and vague - it didn't make the spine tingle with its realism. Depth, as detail, to some degree remained in the domain of 'intellectual perception'. Some of this may be due, as suggested elsewhere, to poor microphone technique - in which case it would be another instance of the difference between 'detailed' and 'realistic' sound - but from what I heard this afternoon comparing normal radio 3 programming a lot of it is due to the encoding (having not been able to make a parallel recording of the Bruckner from FM I can't say.)
    I have been surprised to read of others' problems with FM - here in rural West Cornwall there are no problems with birdies or interference, and the noise floor is superbly low. On the other hand, we do not get as broad a broadband as those in the capital, so the givens are different. With a properly functioning FM service I personally will trade some compression (though I wish it were more sensitively done at times) for superior spatial depth and timbre. Maybe I am particularly intolerant of lossy codecs: I remember years ago finding the first minidisc machines a joke, and thirty seconds of radio 3 on an expensive DAB tuner were enough to terminate any interest in that medium.
    So? More of the same, but next time with lossless encoding - presumably flac - please. Now that might really be something! For now I must stop this and concentrate on tonight's Prom - most of it on FM I fancy.

  • Comment number 58.

    Concentrating on the comparison requested here is my list, based on the firxt half of this evening's BBC SO programme:-

    Advantages of 320kbs AAC XHQ compared to 192 kbs AAC: listening on Sennheiser HD25 headphones + Creative Audigy 2 ZS outboard interface.

    Announcer voice cleaner, much less slurring of sibilants
    Piano warmer
    Reverb warmer
    Pianissimo very much warmer and clearer
    Orchestral fortissimo more attack and more weight (noticeable in Schumann Introduction and Allegro appassionato)
    Very firm and consistent balance between piano and orchestra.
    Non-musical noises are less prominent
    Feeling of greater involvement - returning to 192 kbs feels as if a veil or gauze is interposed

    Comparison on speakers but at a lower level (neighbours) there's still a more "relaxed" and "open" feel to the XHQ sound compared to the 192 kbs AAC

    Apologies if some of these don't make objective sense - I am reporting subjective impressions noted during the performances.

  • Comment number 59.

    This was the first time that I have listened to R3 via computer. Of course even given my limited technical resources it was perfectly clear that the XHQ stream is vastly superior to the flat sounding standard i-player version.

    My reason for posting is to express my doubts about the validity of the survey. I was asked to rank DAB, FM and the XHQ stream. I answered that my experience is that both DAB and FM are better. Why? Because my DAB and FM listening is via close to start of the art audio equipment, both digital and analogue. My listening to the stream took place via a cheapo OEM soundcard, 40 quids worth of Logitech desktop audio amplification and some very good headphones. Not surprising that I prefer the old media given the technical superiority of the rest of the listening chain. However had I been able to send the XHQ stream from PC to my big audio system then I expect that I may have found it preferable.

    The point is that the survey gives no chance to compare like with like and takes no account of the compete listening chain. Thus my response whilst compliant with the survey is also misleading.

    BTW, as mentioned elsewhere : why do the announcer and the interval material need to be broadcast at such high levels forcing me to rush across the room to turn down a voice that is (subjectively) about 10dB louder than the mean level of the orchestra (on all media)?

  • Comment number 60.

    I would disagree with fogou about the comparison with FM. I wonder if he would get better results from 320 AAC if he had a separate DAC? I use a Magnum Dynalab MD100 for FM, and the comparisons with the AAC streams are telling. The R3 internet stream is fed from my Macbook via Wireworld glass fibre optical cable to a Cambridge DacMagic. Dynamic range apart (seriously compromised on FM), the 192 kbps stream is almost identical to the Magnum - just a very slight rounding off of detail and ambience, and a marginally less rich timbre on winds and strings etc. I don't in all honesty find those slight losses musically significant. But it usually sounds a little fresher and bigger in scale too - the image of the orchestra a little larger and more present - and the much wider dynamic range seals my preference for it. At 320, I can detect little or no difference between the Magnum FM and the Mac/DacMagic combo, in terms of detail and ambience; but now the AAC stream has greater presence, with a greater image scale and very good dynamic range. It is also easier on the ear at high levels, a characteristic one always notices with higher-rate digital, say when comparing lossless files to 24/96. It does indeed remind one of how R3FM was once allowed to sound!

    Well done to the BBC and Rupert Brun's team - I'm sure we all hope for R3 to run at 320kbps permanently. Or if this isn't technically possible (why not?), perhaps use it for BBCSO at the Barbican and so on.
    Jayne Lee Wilson, Liverpool UK

  • Comment number 61.

    I don't know if this is a general experience, but the stream cut out here at 20:47 and had to be restarted. I was out of the room when it cut out so don't know if it was just a transient loss or whether it was out for some time.

  • Comment number 62.

    I was very pleased and excited to hear about this experiment, and I was looking forward to hearing the Proms live in 320 kbs quality though my hi-fi system which includes an internet radio streamer with a high-quality DAC. Sadly it hasn't turned out that way, as the XHQ stream seems to be available only as an rtmp: Flash stream and my internet radio streamer requires an http: stream. For normal listening I use the 192 kbs Radio 3 http: WMA live stream, but I can't find an http: version of the XHQ stream. I tried playing the XHQ Flash live stream on my laptop and connecting my laptop's headphone output to my hi-fi, but the benefits of XHQ were negated by the poor quality of my laptop's audio output. The only A/B comparison I have been able to do was laptop/320kbs/hi-fi against laptop/192kbs/hi-fi, and the 320kbs stream did sound better.

  • Comment number 63.

    Just to add my system details to my post above: the Magnum DynalabMD100 FM tuner, and the Macbook/ Cambridge DacMagic, are connected to ATC Pre/Power amplifiers and Harbeth Compact 7 Speakers. This is a very truthful and revealing system, with a clear view of the characteristics of the source. It's important in view of some comments about DAB and FM. I honestly think anyone comparing the various transmission chains in a hifi system of any reasonable quality would prefer the AAC streams, especially at 320kbps.
    Jayne Lee Wilson, Liverpool UK

  • Comment number 64.

    Welcome the subtle pull back of the applause at the end of the Mozart Symphony No. 40 to allow the announcer to be balanced just a bit a bit lower, as we have all requested.

    No break here since establishing the feed at about 1930. Breaks like that are a problem with internet feeds, which often don't automatically re-establish. If it wasn't due to your screen-saver or some other power saving kicking in then internet congestion is usually blamed, which is all but impossble to trace.

  • Comment number 65.

    Just to add my thanks for, and congratulations on, the successfully tweaking of the dynamics in the chain. Not only tonight's Prom (the first half of which I managed to get uninterrupted), but this afternoon repeat of Prom 63 and the Lunchtime concert before it showed much more finely graded dynamics. I do so hope the outstanding success of this experiment leads to a more permanent offering. I eagerly await tomorrow night's Monteverdi.

  • Comment number 66.

    with regard to JLW's question - yes, I would love to try the digital output decoded via the Quad 99 CD-P downstairs, and it might indeed modify my preferences, but that would require an optical cable of about forty feet so unfortunately it's not going to happen, which is most frustrating. But my quad fm4 is presumably not as good as the magnum dynalab tuner either. As PAR so wisely pointed out, "the survey gives no chance to compare like with like and takes no account of the compete listening chain", which is why it is useful to know what people are using. The speakers I used to listen to the Bruckner are also 'BBC school' like JLW's, plus a subwoofer. The Terratec soundcard I fitted to the PC had a couple of zeros after the decimal point for every test of harmonic and intermodulation distortion, got "excellent" ratings in a review - for what it's worth - particularly for music reproduction through a hi-fi system and was just about state of the art at the time. I don't know what its low level linearity performance is however, which could certainly be relevant. I would expect the Quad's upsampling DAC's to be better.

    If I compare a CD played on the computer to a Sony ES CD player's version, the computer gives a slightly brighter account but the imaging remains good, I am still well aware of the interaction between instruments and acoustic in a very 'live' and finely textured presentation. I do prefer the Sony, but then I would hope so, but the differences are nowhere near as great as when comparing the different accounts of Jennifer Pike's recital this afternoon. (In storing music on computer I have come to use only flac, despite the file size, which is perhaps significant.)

    What I find very interesting is that JLW's "just a very slight rounding off of detail and ambience, and a marginally less rich timbre on winds and strings etc." for 192 v FM exactly matches my perceived differences for 320 v FM, except for 'very slight' and 'marginally'. It could be the equipment or it could be the listener - the two sounds were essentially very similar it's true, but for me the differences were very important. Clearly other contributors are far more bothered by reductions in dynamic range on FM than I am, and I am probably particularly bothered by losses concerning the interaction instrument - space.

  • Comment number 67.

    My thanks to fogou for continuing this enjoyable discussion. In reply, I would say it is both the equipment and the listener that results in the perceived differences. I too love spaciousness in reproduced classical music, and the DacMagic is quite exceptional here, especially through the optical input (disclaimer! I have no commercial/vested interest in the Cambridge company!) So with the AAC stream it does really contribute to that sense of presence and realism which was formerly, in broadcast and webcasts, the preserve of FM. As for this listener, I attend RLPO concerts fairly often, and inevitably an FM relay seems disappointing after that, or compared to CD, especially with larger orchestral works. A Bruckner Seventh without climaxes?! But even a Haydn symphony can seem to lack punch and drive on FM these days. I would urge fogou to try to arrange a feed for AAC into his 99cdp, or any separate DAC. He sounds like the sort of listener who would appreciate the result.
    Incidentally, can I ask Rupert Brun and the team if the XHQ stream will still be available for next week's Prom repeats, ie. after the end of the season proper?
    Jayne Lee Wilson, Liverpool UK

  • Comment number 68.

    Ah well, JLW, compared to live sound yes indeed FM is disappointing, CD is disappointing, LPs were disappointing, stereo is disappointing ... in dynamic range, localisation, that paradoxically banal non-quality of total absence of unnatural artefacts and, particularly with big orchestral forces, what subjectively I experience as a kind of delicate fluffiness to the sound - the grid we use to reproduce sound just seems far too coarse in comparison. Unfortunately tomorrow and Saturday night I'm not going to be disappointed because I'll be at chamber music concerts here, not listening to the radio, so that pretty much puts paid to somehow arranging a feed to the Quad CD-P, unless of course your final question gets an affirmative answer.

  • Comment number 69.

    Just a couple of follow ups to points raised by other contributors:

    fogou, If in the future you want to link music from your PC to the digital input of your audio system there are various solutions that don't involve festooning your house with cable. I use Logitech's Squeezebox (actually the Squeezebox Duet) but there are other products on the market. The Squeezebox has a software plugin to enable you to access the XHQ and the iPlayer streams. (It is also very effective indeed when used to play music files, say CDs transferred to a hard drive. Many people report that the result is comparable to very good CD players.)

    johnlondonw14,
    I'm sure that some of the breaks in the XHQ stream are due to internet issues but quite a number of times I've experienced the XHQ stream stopping and been unable to restart it with the URL that the stream had been using. It has only been possible to restart the stream when a new (different) URL has been obtained. This has been when accessing the stream through Squeezebox, etc - not directly through the BBC website, which might behave differently. That is why I suspect that the URL is changed periodically - though I might well be talking nonsense. (I frequently do.)

  • Comment number 70.

    re: routine disconnections at 4pm

    I haven't experienced that here. Ocasional random disconnections yes. And some diconnections earlier this week when the BBC were clearly adjusting the system.

    I suspect internet provider issues: 4-6pm is notorious for a peak demand on internet.

    Lack of reliable connectiuons would be an indicator against a long term 320kbps facility unless an alternative is available at lower bit rate.

  • Comment number 71.

    More comments to follow, but here is a plea to Rupert to allow the XHQ bit-stream to continue until at least next Monday. Why? Because I could capture again the BPO/Rattle Mahler 1 afternoon repeat without the limiting problem.

    Has my request been made in time?

  • Comment number 72.

    We have just started the Monteverdi Vespers, and the level is much higher than when I last listened on Wednesday. Already, it is hitting -4dB at points during the opening chorus, so it must be being clipped. What a waste. It was perfect for the Bruckner symphony. Why put it up? If it is like this for Monteverdi, heaven knows what clipping there will be for the Last Night. I have not tried the 192 stream to see if it is any different. I must say, this is disappointing.

  • Comment number 73.

    Listening on my Beyer DT770's. Sounds superb! Startling dynamic range and realism. As some others have said, can't we always have R3 at 320, please? BTW, what rumbled at 20:30? Short-term aircon, or Concorde flying outside?

  • Comment number 74.

    I hadn't been listening to tonight's Prom but, having seen comment 72 - cantobel, I did a quick check on the XHQ stream just after 8 pm. Within the minute the level on both channels had hit - 4 dB. However I haven't seen another instance in the following 30 minutes. Maybe the level at the RAH was a little too hot and they were caught out. Choral content can be very 'peaky'.

  • Comment number 75.

    Now listening to the XHQ stream through my ancient Quad 44/405 gear + Rogers Studio 1 Speakers.
    The difference between the stream and FM via my FM4 is telling.

    The XHQ stream is sublime! And no dropouts ..yet.

    Much more please!

  • Comment number 76.

    After those brief moments of strain early on as noted by cantobel and rmgalley, I thought the Monteverdi was outstandingly pure, natural and beautiful, and yes, spacious - a great example of what the higher-rate AAC stream can do. Earlier in the season I abandoned some choral concerts because the 192kbps stream proved rather hard on the ears, especially the higher pitches at higher levels. The Vespers is not at all my usual fare, but I stayed to the end, and the sound quality made the difference - it drew me in and held me. A great occasion!

    i have not suffered any dropouts or connection breaks on XHQ. Last season (2009 Proms) I was getting 2-3 dropouts per night, sometimes a few seconds long. This was because the ISP, BT Option 1, was reducing iPlayer speeds during peak times (presumably it still does?); the iPlayer speedchecker often gave readings of well below 1000kbps! Upgrading to Option 2 cured the problem completely and the connection stabilised mid-evening to 5-6 Mbps. So if you're on a budget option check with your ISP. On the Mac, I save the iPlayer R3 and XHQ Proms pages as a web archive, and return to that each time (by clicking twice on the file). Works for me!
    Still wondering (Rupert?) if any of next weeks' Proms repeats will be XHQ...
    Jayne Lee Wilson, Liverpool UK

  • Comment number 77.

    Rupert , I already have some suggestions for you to consider but first a summary of where we are at present. With so many different platforms for the delivery of Radio 3 Proms evaluation has become quite a protracted process. For Radio 3 we have:

    • Analogue FM
    • Digital satellite (Freesat/Sky) mp2 coded at 192 kbps
    • Terrestrial digital (Freeview) mp2 coded at 192 kbps
    • DAB mp2 coded for the most part at 192 kbps (except when R5SX is on air when it is reduced to 160 kbps)
    • The various internet streams (I won't consider here the WMA streams) …

    The aac streams possible are more than you might think:

    • The normal ‘Listen Live’ 192 kbps coded stream of the live concert (live/live)
    • The XHQ 320 kbps stream of the live concert (also live/live)
    • The afternoon ‘Listen Live’ Proms repeat (usually the following week) available as a 192 kbps stream (repeat/live)
    • The afternoon repeat with the XHQ 320 kbps stream (repeat/live)
    plus ....
    • The iPlayer ‘Listen Again’ of the live broadcast at 192 kbps (although this stream purports to be at 128 kbps I have found, on careful checking, the amount of data downloaded is consistent with a bit-rate of 192 kbps)
    • The iPlayer ‘Listen Again’ of the afternoon repeat broadcast (probably also at 192 kbps, but I haven’t checked this)

    I would be very easy to refer to any one of these without being sufficiently precise as to which feed/stream was being commented on. This is particularly relevant since the changes to the online streams on Wednesday 8 September have meant the repeat Proms, broadcast in the afternoons, have a different audio signature to those originally broadcast on both the normal and XHQ streams.

    The only analogue Proms source is FM, and while the basic sound quality is very good, it is now marred by heavy dynamic range compression. No doubt this is to make it more listenable in fringe reception areas or to compete in the station loudness stakes.

    All the mp2 sources sound basically the same. The exception to this is DAB. A lower bit-rate is sometime used and the reduction in audio quality escalates at anything below 192 kbps. Also the audio is more easily impaired if the reception path is less than ideal and the dreaded ‘underwater burbling sound’ becomes manifest as the error protection methodology is not up to the task.

    The Online Streams:

    It used to be the case that the ‘Listen Again’ iPlayer stream was much inferior to the ‘Listen Live’ one. There used to be very noticeable dynamic range compression. In addition some ‘Listen Again’ repeats I listened to were badly marred by too high an audio level with limiters ducking in response to high level transients. These left a momentary hole in the music before they released. I am happy to report, since Wednesday this seems no longer the case. The Radio 3 ‘Listen Again’ is now without compression and sounds much better for it. I do have one quibble. The system limiters are still set too low. Here anything higher than -6 dB of FS is flattened off (but without any evidence of ducking). I don’t know if the R3 ‘Listen Again’ stream was ever at 128 kbps but, although it is still flagged at that bit-rate, it is actually at 192 kbps. The only other station I checked was Radio 4 which is flagged as 128 kbps and is at that bit-rate.

    I suspect we would find the live/live and repeat/live ‘Listen Live’ streams to be identical now. Unfortunately the XHQ experiment will end before this can be established. When the experiment began the levels of, and lack of limiting on, the mp2 feeds and the normal 192 aac ‘Listen Live’ stream were all identical. Prior to 8 September the XHQ stream was 6.9 dB higher in level and peaks were regularly hitting the limiters. Since that date the 192 and 320 aac streams have parted company with the mp2 feeds in that, while being identical to each other, are now both 1.9 dB higher than the mp2 feeds. This afternoon I only checked the XHQ ‘repeat/live’ of Saturday’s BPO/Rattle Prom, but the highest level in the Schoenberg was still hitting the limiters. I was using the digital satellite feed as a reference.

    I would question the setting of these limiters at -4 dB of FS. This afternoon the peaks should have gone to – 3 dB. Rupert, you correctly state “We also have to consider what happens to the audio further downstream - every process we apply to a digital signal has the potential to create overshoots which can become clipped further along the chain”. It is absolutely true coding artefacts of lossy compression systems create overshoots. If the audio were at, or near, FS the decoded audio could have clipped content. However these overshoots are only significant when high levels of lossy compression and very low bit-rates are used. The overshoots on 192 or 320 kbps aac audio are negligible and I should have thought limiting at - 0.5 dB FS would have provided sufficient protection.

    Of course ‘what happens to the audio further downstream’ could relate to the broadcast material being transcoded to a 48 kbps mp3 file but then the normal coding artefacts at that bit-rate would be more significant than any clipped content. With the changes made on Wednesday you increased the levels by 1.9 dB with respect to the mp2 feeds. That utilises more of the dynamic range of the 16-bit system but the thresholds of the limiters need to be changed too. Would you consider raising the threshold to -2 dB or higher? Also the Radio 3 ‘Listen Again’ is currently set even lower and should be set to the same threshold as the ‘Listen Live’?

    I am still struggling to make valid direct listening comparison between the 192 and 320 kbps aac streams using identical equipment (of known merit) at my end. Switching between the two streams in the web browser necessarily creates breaks while buffering takes place. Switching between tabs, muting and un-muting the iPlayer, is not much better. More to follow ...

  • Comment number 78.

    Thank you everyone for the considered responses. First the bad news - I am afraid the experiment will end after the Last Night live broadcast.

    For the good news, I agree there may be room for me to raise the level of the limiter immediately before the codec further without the codec clipping and I will investigate this carefully after the experiment is over. For the purpose of this trial I didn't want to go too far because I wanted to match the levels of the main 192 stream and the experimental one, so that you could do a fair comparison of the two. It would not be appropriate for me to make radical changes to the main stream without proper listening tests and consideration and there wasn’t time to do that, so the present settings are a cautious compromise which I shall revisit when there’s time for a more thorough examination.

    The slightly hot start to the Monteverdi was not because of any changes to the experimental feed, I’ve not made any further changes since the adjustments of the 8th.

    Enjoy the “Last Night”!

    Rupert Brun, Head of Technology for BBC Audio & Music.

  • Comment number 79.

    Well done Rupert Brun and all at R&D...Radio 3-320kb/s AAC...brilliant dynamic range.used Bose 3 ANC Headphones.Prom 69 Beethoven 5th.just like standing at front of Arena..all the coughers included..look forward to having XHQ as permanent feature ..soon

  • Comment number 80.

    Rupert

    As the first 'listener' to post to this thread, can I add my heartfelt thanks to you and your team, not only for the feed, but perhaps almost as importantly for the feedback.

    I'm sure many of us will look back on this week and especially this Last Night with some nostalgia and perhaps a little pride.

    'We Were There' [Sorry Max!].

    Just a couple of extra questions - what if we all slip you a fiver each, could you just be persuaded to forget to flip the switch on your way home tonight?

    Also it is common practice on the BBC to offer support via a Help Line to those who have been adversely affected by the Corporation's output. Will there be a service available to those of us who have become addicted to the 'high octane drug' you have been supplying to us since last Friday Week?



  • Comment number 81.

    Lovely clear, open sound again tonight. This is addictive!

    Well done Rupert and team!

  • Comment number 82.

    Yes, magnificent tonight, though the coughers and sneezers are out in force. Blest Pair of Sirens really is demonstration quality. Well done and many thanks. Only Mr Rafferty's voice managed to hit the limiter.

    What chance of delivering this stream in the future over satellite, which is surely a more stable channel than internet? A few of the best continental classical stations (Bayern 4 Klassik and Austria O1) are already transmitting in Dolby that way. Of course, if the BBC could relinquish just one of the 101 versions of BBC1 on satellite, all exactly the same for most of the time, there would be sufficient bandwidth to transmit all four main radio networks in completely uncompressed PCM.

    Let's hope we don't have to wait too long until this experiment is made permanent.

  • Comment number 83.

    I have missed most of the 320 kbps experiment through being away on holiday until today. I would not usually listen to the last night of the proms but I am doing so this evening in order to try the new high quality stream. I have my mac digital output feeding into a Naim DAC, and a Naim tuner running the programme in FM as well. The 320 is good, the best digital yet, but switching over to FM brings a much more musical sound, more bass, richness, depth and warmth. But this music is horrible, I must switch off - but continue in 320 for streaming please.

    d

  • Comment number 84.

    I've spent most of this week searching for a way to convert the Flash/rtmp live stream into a form that my music streamer can consume. Finally today I managed to do this using a multi-stage process that unfortunately requires two additional restreaming steps on my laptop (one using a lossy codec) before sending the stream to my music server. I'm pleased to say that the laptop-restreamed 320 kbs Flash/rtmp stream sounds slightly better on my hi-fi system than the "straight" 192 kbs WMA stream, despite the laptop-restreaming issues. I have also sent the normal 192 kbs Flash/rtmp stream through the same laptop-restreaming process to get another comparison point, and this sounded worse than either of the other streams. I expect a 320kbs WMA stream would sound even better on my equipment, and I'm looking forward to being able to hear that.

  • Comment number 85.

    Well I suppose there was bound to be someone who prefers the Optimoded, frequency constricted, birdie ridden, relatively high noise-floor sound of FM. Strangely, both my ears and spectrum analysis show a rather better bass response from the XHQ stream than from FM, but I guess it comes sown to what some people are used to. Oh, by the way, Radio 3 FM has not been analogue from mic. to transmitter for years. You are listening to music transferred, via a good digital codec, from the RAH to the Radio 3 Optimod set-up and then on to the FM stereo multiplexing (with its relatively limited channel separation) and then on to the transmitter and your receiver. Please don't kid yourself that FM is anything other than a poor elderly relative of the 320kbps aac stream. More 'musical' than 192kbps DAB (though with poorer dynamic and frequency range than the latter) but just a squashed version of what has been on offer via the XHQ stream for the past couple of evenings (once Rupert and his team had sorted the dynamics).

    I will miss the convenience and low cost receivability of FM when it goes in a few years time, but not its relatively poor audio quality (compared to the XHQ stream, that is).

    Hopefully we have heard the future of Radio 3 Internet audio, and it works.

  • Comment number 86.

    I found out about the 320kbs experiment too late to take full advantage of it, but I listened to the Last Night of the Proms, alternating between DAB and the AAC stream. Unfortinately I had not time to provide my PC with a decent D/A convertr, and so had to work with a converter that is inferior compared with that in my Roberts DAB receiver. The results made an interesting comparison, however. I listened to both sources using both Koss headphones and Mission speakers driven by Cyrus amplifier. Both sources held advantages in my own trial. Listening to the AAC stream I found the high frequency peaks were confused and noticeably less bright than the DAB source, however I have little doubt that this was a limitation in the D/A converter I used. Listening to DAB gave a more rewarding experience in that respect. However there were other respects in which the AAC stream was definitley superior, not least being the clarity and spacial nature of the sound stage. There was much more sense of "being there". The dynamic range was also noticeably improved with much less obvious compression except on a couple of peaks towards the very end of the Last Night when the audience and orchestra was at their noisiest. It must be very difficult to broadcast such an event in any format without at least some protective compression at some point, however its general absence in the AAC stream was very enjoyable.
    I do listen to internet streams quite a bit, but do not usually expect or demand extreme high quality from them, largely beacuse either it's not generally available or I'm not listening seriously. For serious listening I usually record direct to MP3 or WAV for later listening on a decent device, so the limitations of the D/A converter currently connected to my PC do not normally intrude. But I'll have to go out and get a much better D/A converter for my PC to get the best benefit from broadcasts such as this; it would be nice to think it would be worthwhile justifed perhaps by more BBC broadcasts such as this. Well done Beeb. Lets have more of this please. This was a Good Thing.

  • Comment number 87.

    Just a quick thank you to whoever decided to leave the facility on throughout last night's "Hear and Now", (or forgot to hit the off switch). I feel sure the composers whose work was included would be very pleased to know that it was available in such high audio quality for those taking the risk of the experiment being terminated mid-programme.

  • Comment number 88.

    It feels a little like after the Lord Mayor's show today. Oh well, back to normality. I can echo Nethersage's comment about 'Hear and Now' last night. For a radio stream the sound was extraordinary. The music of course was at polar opposites to the Last Night frolics!
    Very many thanks to Rupert and his team for the past week. The memory will linger. We await future developments with impatience........

  • Comment number 89.

    Now this is all very interesting;)..

    Consider that on the Proms stage the performers are highly talented musicians, each and every one of them. After all not just any olde fiddle scraper gets into the BBC SO do they?. Do any of the said musicians use cheap "knock off" Strad copies?. Do any of the conductors not care about the musical balance, or their interpretation of the composers intentions? I rather suspect not;!.

    So all in all they strive to produce a musical performance of the highest standard. World class many will say..

    Now this costs a lot of money to do. A lot of licence payers money too and of course not all the licence payers could fit into the Albert Hall let alone for many, get there. So we have this wondrous system where people a long way off can hear and sometimes see whats going on there;).

    Now logically, why should the BBC entrusted to do this, of all people treat this performance and final sound thus produced with some contempt in that they on FM radio compress it, and distort the dynamics of same?.

    On Digital Terrestrial radio, use an outdated codec MP2 at lower than the bit rates intended for its use and in doing so introduce information that wasn't there, and throw away some of the information that was there in this system.

    So now a step forward and a realization that this can now be done better. All credit to Rupert and his colleagues for thus doing so.

    However is this the right way forward?. Can he answer this point please?. In other European counties such stations as France Musique manages 256 K on Satellite. Bayern Klassick 4 manages a whopping 320 K and over 400 for surround sound!.

    Why does the BBC not up the rates on satellite?. After all a lot of people already have one of these in their living room and even if the audio output stages of said box aren't that good many now have SPDIF outputs thereon and can take that to a high performance external DAC unit .. job done:) You don't even have to have a Sky sub either, there are plenty of Digital Free to Air boxes around. I've got one. Works very well for 't telly too!.

    Now not everyone has a decent BB connection. I have a 10 meg VM service here backed up by ADSL and even that was dropping out, due I presume to demands on the network/s. Not everyone has a PC in their living room either and in general most of these are noisy and need a display screen to see what your doing. Not everyone has access to a high speed BB service, even within Three miles of Cambridge where I live some think themselves lucky to get anything over a Meg reliably!. Some use a BB dongle now via the mobile Telcos. Some are seeing caps placed in their downloads unless they pay for industrial grade services. So all in all better, but is it that -good- an answer.

    So a couple of questions please Rupert.

    FM can be very good given a decent tuner and most important a decent well installed aerial. Now why do we still have to suffer Audio processing on the Proms, for that the evening output of Radio 3.
    OK audio processors have their uses for some services at some times of the day but in the evenings for serious listening can you not either switch them out or down to mimimal settings?.

    Also why cannot, as suggested, we have some "upped" rates on satellite?. This is a very simple way to better the broadcast quality, there's plenty of bandwidth and capacity to do this. Why not a 256 K service or even one as good as the Germans manage?. Its rather ironic that the TV sound is at 265 K so why do we have to suffer radio at a lesser rate?. Why sometimes as for Radio 4 even in Mono on T-DAB let alone that joke of intensity stereo.

    Why can't we have a fit for service radio broadcast system in the 21st century?.

    Please?..

  • Comment number 90.

    I'm very pleased to be able to get on-line via stereo headphones sound quality from the Proms to better the best of the old analogue FM broadcast, and indeed to match the sound quality of my best CDs. Only actually being there (as I did as an arena Promenader for many years) would be better. Well done BBC! Hope you are able to deal with the technical issues discussed in these posts, so that we can have 320kbps AAC as routine on Radio 3 in time for Proms 2011.

  • Comment number 91.

    Rod,

    "Hope you are able to deal with the technical issues discussed in these posts, so that we can have 320kbps AAC as routine on Radio 3 in time for Proms 2011."

    Please, not just for the Proms but for Radio 3 24/7!

  • Comment number 92.

    320kbps AAC is undoubtedly a godsend for Audiophiles who will be prepared to go to the effort of arranging to link a PC with a soundcard up to their HiFi via SPDIF, or buying a standalone internet audio player capable of receiving 320kbps AAC. (for the latter the BBC will have to pay more consideration to advertising URLs rather than using Adobe Flash for the front end).

    However, we must be clear; a general user with a standard PC with integrated sound card and sub-standard speakers is clearly not going to benefit from such a service.

    The BBC is going to have to do a cost / benefit analysis on this. If it became a guaranteed ongoing service I'm sure that it would give a great imputus to the HiFi industry to produce internet players at lower prices. It would also do a lot to quell criticism of the other distribution platforms, as those who are unhappy at present will also generally have the ability to make use of the HiFi service.

    But the internet is, by design, not best suited for one to many broadcasting (at least until ISPs embrace multicasting). So this service will have to survive on the basis that it is satisfying a niche.

  • Comment number 93.

    Sadly I had to work on Friday night but managed to set up a scheduled recording of the Monteverdi Vespers as a lossless original. I had a quick look at the BBC4 "watch again" version to get my bearings, then listened to the XHQ stream. What a sublime performance and what wonderful sound - like going into the sunshine.

    Since I stumbled on XHQ for Prom 66 last Saturday I've been hooked, checking this page every day, following the resolution of the technical challenges and adding Spectre to my iMac so that I could see what people were talking about. For example, I could hear that the Rite of Spring was limited, whereas the Monteverdi Vespers stream seemed pretty well perfect.

    A big thank you to Rupert Brun and all of his BBC colleagues for giving us a wonderful week of Radio 3, especially the live broadcasts but also the daytime programmes. I'm definitely going to miss XHQ but hope that the undoubted success of this experiment will lead to greater things.

  • Comment number 94.

    There are many questions raised by this experiment, probably not all intended. However I agree with all the comments made by #89 Tom Sayer.

    There isn’t much wrong with the mp2 codec if it is used as intended. It was a state of the art development in the late 1980’s. But with the then complexity of decoders and the lack of available processing power by today’s standards, it meant a rather simple perceptual coder was all that could be achieved. The problem now, as processing power is cheap and readily available, the mp2 codec is effectively obsolete. The German satellite radio stations nearly all use mp2 at 320 kbps and sound superb but there are more efficient, more cost effective and, crucially, higher quality codecs available. The aac codec that is now used for the online streams and the XHQ experiment is an example.

    Back whenever it was, when the BBC leased additional satellite transponder capacity from SES-ASTRA for all the national and regional variants of BBC-1, I and others put forward the suggestion a 256 kbps bit-rate could be deployed on all the radio services. I can’t remember exactly but it would have utilised about 3% of the leased transponder capacity.

    The BBC were then, and still are, promoting DAB. The limited radio spectrum, and the choice to broadcast so many stations, means the bit-rates are much too low. So, as noted, the audio quality is mediocre and occasionally stereo programmes on Radio 4 have to be broadcast in mono on DAB. This restriction does not apply to digital satellite where the TV services, including the Red Button channels, hog nearly all the capacity. A conclusion was reached back then that the BBC were unwilling to up the bit-rate on the satellite platform because the public would be able to make a direct comparison with DAB and come to appreciate how dire it sounded. I suspect there are still likely to be quarters in the BBC who remain of this opinion. So this is why it is so welcome we appear to have a champion in the BBC, such as Rupert, who actually seems to care about audio quality.

    There is no technical reason why the aac streams couldn’t be up-linked to the satellite for direct reception in the home. This would circumvent any reliability issues with the internet.

    It used to be the case radio 3 FM was not dynamic range compressed in the evenings after ‘drive time’. This is certainly not so now. I have a PCM-701 digital recording from FM of the Mahler 3/Berlin Phil/Abbado Royal Festival Hall broadcast of 11 October 1999 – with absolutely no compression and very wide dynamic range. The concert was also recorded by DG for later release on CD. Some commentators have even complained about the dynamic range on this! But the broadcast dynamic range and that on the CD and exactly match and both sound wonderful. The point - FM could still sound superb.

    Dynamic range compression has no place in digital systems and I wish to thank Rupert for removing it from the 'Listen Live' daytime stream of Radio 3 and also from the 'Listen Again' stream which sounds so much better now.

  • Comment number 95.

    Just before people get carried away with the idea of making satellite the platform for a high quality feed, may I politely point out that only a minority of households actually have a satellite dish!!!

    Also, I believe that dynamic compression, performed manually or automatically, has always been used on FM. Perhaps less so in the evenings, but present never-the-less. (Admittedly the Optimod system or similar, that the BBC has been using on FM for many years, is particularly pernicious.)

  • Comment number 96.

    @95

    The same could arguably be said of computers and DAB radios. Not every household owns one or the other or even both!

    The BBC is a multi-media broadcaster (TV analog & digital, Radio analog & digital, terrestrial, satellite, internet...)

    It seems to me that each and every user, wherever situated, should be able to enjoy the output at its highest quality whatever the media. Broadcasting radio at 320 kbps on the satellite would be an absolute delight and would put British broadcasting on a par with France and Germany.

    Thanks to Rupert and his team and also a special thanks to the BBC for having made the EHQ stream available to us non UK residents.

  • Comment number 97.

    #95 johnb

    Until such time as every home is served by fibre, satellite is currently the only platform able to support a multiplicity of high quality feeds. The DAB multiplexes are full. The BBC’s Freeview multiplex is full.

    We have been afforded the luxury of a high quality audio stream feed on just Radio 3. The nature of the internet is such the consistency of the flow of data is outside the control of the user. Even breaks of a few milliseconds once every hour can be disturbing.

    Nevertheless I sympathise with those unable to ‘see’ the satellite or live in a listed building where planning permission would not be granted and satellite is not an option.

  • Comment number 98.

    Comparison with the better continental stations is interesting. It's true that many of the German (and other) classical stations broadcast on satellite in mp2 at higher bitrates than 192. However, the bitrate isn't everything, as we've seen in this discussion of the Proms experiment. Bayern 4 Klassik, which is often held up as the shining example, has fairly recently added a limiter on its mp2 stream. It doesn't seem to be particularly audible, but it is very evident if you look at the file. Only their Dolby satellite stream is not limited, and that is sometimes stereo, sometimes surround. Similarly, Dutch classical radio is at 384kbps mp2, but quite heavily limited.

    Ideally, we would prefer no limiting and a higher bitrate from the BBC. But the easiest way to improve the listening experience, which costs not one penny, is to raise the limiter level and to set the audio level so that the signal never reaches the limit. That really should not be so difficult.

    As for availability of satellite, that really is the form of signal that is available to the greatest section of the population - far greater than fast broadband, or good quality FM or DAB signals. No subscription is required to receive it, and the equipment to receive it is cheap.

    The BBC multiplexes on Astra are full only because the BBC chooses to broadcast every regional version of BBC1 24 hours a day - and each of those has its audio at a superior bitrate to any of the BBC's national radio stations. 95 per cent of the time, they are all broadcasting exactly the same content. Let me just list them, in case anyone feels a burning desire to watch one of them:

    BBC1 London
    BBC1 Northern Ireland
    BBC1 West Midlands
    BBC1 North West
    BBC1 East Yorkshire & Lincolnshire
    BBC1 Yorkshire & North Midlands
    BBC1 East Midlands
    BBC1 East
    BBC1 Wales
    BBC1 Scotland
    BBC1 West
    BBC1 South East
    BBC1 North East & Cumbria
    BBC1 Oxford
    BBC1 Cambridge
    BBC1 Channel Islands


    Add to that 7 "red button" TV channels, at least two of which are hardly ever used and just broadcast a test signal for most of the time. It's a completely absurd allocation of resources.

  • Comment number 99.

    This comment was removed because the moderators found it broke the house rules. Explain.

  • Comment number 100.

    Since the experiment is now over and we seem to drifting off topic I'm going to close this post for new comments.

    Thanks

 

Page 1 of 2

More from this blog...

BBC iD

Sign in

BBC navigation

BBC © 2014 The BBC is not responsible for the content of external sites. Read more.

This page is best viewed in an up-to-date web browser with style sheets (CSS) enabled. While you will be able to view the content of this page in your current browser, you will not be able to get the full visual experience. Please consider upgrading your browser software or enabling style sheets (CSS) if you are able to do so.