BBC Blog Network

User Profile

Viewer00

Listed below are comments made by Viewer00 between Friday, 30th January 2009 and Monday, 2nd March 2009.

«Newer
Older»

BBC HD Update: DOG Patrol 4:46pm on 02 Mar 2009 There's plenty of music in Natures Great Events. Some I know would like to turn it off, but for me it adds to the visuals. Also there are some great subwoofer moments which pep up the dramatic impact. Is there really not enough drama and music in this programme to alert the dog patrol of a needed removals duty?
A Christmas Present from the HD Channel! 12:27pm on 06 Feb 2009 Dear Andy,

Thank you for your continuing interest in resolving the BBC HD delayed sound issue.

I confirm that the photodiode was picking up the top Sync Flash Bar for the following Sync Flash Bar measurement results.
On 30/01/09 at 11:50 AM the sound starts 70ms after the top bar is lit.
http://myweb.tiscali.co.uk/luculent/Pictures/SyncTest0.PNG
On 02/02/09 at 4:50 PM the sound starts 50ms after the top bar is lit.

Also, when the photodiode was picking up the Measurement Bar as it passes the zero calibration mark, I obtained the following Measurement Bar results.
On 30/01/09 at 11:50 AM the sound starts 40ms after the bar passes the zero mark.
On 02/02/09 at 4:50 PM the sound starts 20ms after the bar passes the zero mark.

Please note that for each test transmission (date and time) I moved the photodiode around to the relevant locations against the TV screen and so captured on the .wav file about 20 results for each test (i.e. 20 clicks for each of the three sync flash bars and the one measurement bar). On analysis, for each test, the results (i.e. the measured delay) showed no noticeable variation.

I understand your concern about the validity of the Measurement Bar results, but for myself I can accept that on 02/02/09 at 4:50 PM, the sound delay was just within the -20ms (on the test signal) we are aiming for.

The only explanation I can suggest for the sound starting 50ms after the top bar is lit on the same occasion (Sync Flash Bar measurement results on 02/02/09 at 4:50 PM), it that Philips may do “enhancements” which “compensate” for psychological factors. (i.e. motion and sound are close because the human brain can predict, whereas, because a bright flash takes time to register in the brain, the sound is delayed to allow the brain time first to register the flash).

As for your setup at home, I’m wondering if you are using a separate audio feed from the Foxsat HD (via an amplifier) to your Mission 770s studios. In which case you might be using the Foxsat Lip Sync control to offset both the display delay of your 50 inch plasma, any delay in the Humax Foxsat, and any delays in the BBC’s HD programme or distribution. Whereas if using the HDMI connection, the Humax Lip Sync control cannot offset delayed sound; it can only add to it.

And you may have noticed there’s another “funny” with the Humax Foxsat Lip Sync Control. Whilst the on screen scale jumps in 5ms quanta, the actual delay jumps in ~50ms (guess - not measured) quanta. If your Foxsat behaves this way, then the best you’ll be able to do is +-25ms (i.e. outside the target).

Please do tell us all about your home setup and how you test/measure lip sync.

Future tests I hope to do are:
1. Tuning away from BBC HD and back again during a test transmission to see whether the sync changes.
2. Using the Foxsat Lip Sync Control during a test transmission to measure the actual incremental delay “quanta” against the scale.

Hopefully others (including tagmclaren) will soon be in a position to use your “Method 2” (or adaptations of it) and share their findings.

P.S. For me, the sound in last night’s BBC HD Hustle, whilst occasionally spot on, in some scenes seemed delayed and in others advanced.
A Christmas Present from the HD Channel! 4:14pm on 05 Feb 2009 Dear Andy

It is not any audio/video editing package I use, it is the Nero Wave Editor. There is no video involved. The photodiode output and the sound are simply recorded and analysed as audio waveforms.

When I conceived the test setup I was concerned about the photodiode response and so ahead of you regarding calibration.

To calibrate my test setup I used a clapperboard. I mounted the photodiode 1mm behind the shutter, and the microphone within 1 cm of the shutter.

Also, to check that there were no time differences occurring in the treatment of the two channels during recording and analysis I swapped over the audio and photodiode inputs.

Both calibration results can be seen here
http://myweb.tiscali.co.uk/luculent/Pictures/SyncCal0.PNG
http://myweb.tiscali.co.uk/luculent/Pictures/SyncCal1.PNG

The calibration results show that as fast as I could shut the clapper, and whichever channel takes whatever signal, the test of the test setup exhibits no detectable timing error.

The rise time on the photodiode trace represents the clapper shutter moving 5mm in 1ms. That’s over 100MPH! At this speed it was impossible to stop shutter bounce, and you can even see the test setup continues to follow this faithfully.

I hope you will take my BBC HD sync test results as an accurate representation of what I experience when watching BBC HD programmes. As I’m using an HDMI connection and have no control with which to reduce sound delays, I’m still hoping that the BBC will work with the receiver and TV manufacturers so that the BBC HD programmes can be watched with sound in sync (at least within your 20ms target).
A Christmas Present from the HD Channel! 4:40pm on 03 Feb 2009 I've put up a screenshot of the Nero Wave Editor window which shows a BBC HD Top Sync Flash Bar measurement I made.
http://myweb.tiscali.co.uk/luculent/Pictures/SyncTest0.PNG
The sound delay is equal to the difference between the "selected" timings shown at the bottom of the window.
This measurement shows that on 30/01/09 at 11:50 AM the sound was 70ms later than the Top Sync Flash Bar with my reception setup [Humax->HDMI->TV].
A Christmas Present from the HD Channel! 5:52pm on 02 Feb 2009 Following on from your reported corrective action, here is my next BBC HD Sync Test Result.
Date: 02/02/09
Time: 4:50 PM

The Setup and Test Method was the same as the test on 30/01/09 at 11:50 AM.

Test Results.

For all results the sound delay was 20ms less than measured on the sync test broadcast on 30/01/09 at 11:50 AM.
[Note: Due to the 100Hz strobed backlight on my TV, all my measurements have to be in 10ms intervals]

Sync Flash Bars.
The top bar was visible. The sound starts 50ms after the top bar is lit.
The sound starts 40ms after the second bar is lit.
The sound also starts 40ms after the middle bar is lit.
The sync flash bars seem to occupy 2 lines (odd and even fields). A second flash (greater brightness seen by the photodiode) is evident, occurring 20ms later than the first.

Measurement Bar.
At the zero point the sound commenced 20ms after the measurement bar passed the photodiode. At the sound late 1 frame point the sound commenced 20ms before the measurement bar passed the photodiode. My visual perception agrees with the test results (i.e. the sync test shows the sound to be late by half a frame).

Comment.
I did not detect any obvious sound delay in the preview footage immediately before the test.
A Christmas Present from the HD Channel! 12:25pm on 02 Feb 2009 Dear Andy

Thank you for your reply to my sync test post and for your work in progressing solutions to the BBC HD sound delay problem.

You ask what the .wav file is for.

The .wav file is a 44100Hz sampled 16bit depth two channel recording. The input for one channel is the TV sound, the input for the other channel is the amplified photodiode signal. This .wav file, together with the Nero Wave Editor software takes the place of the dual beam oscilloscope as described in your blog Method 2 “How do you use the signal to check audio/video A/V sync?”.

I did measure AV sync from a screen where audio and video are via the HDMI connection. You say the only accurate way is to use a microphone against one of the speakers. I was using a direct electrical connection from the TV’s headphone socket. I feel this should also result in a valid measurement as I expect the headphone electrical signal to be synchronised with audio sound from the speakers.

The AV sync via the HDMI connection from the Foxsat is adjustable, but the sound can only be further delayed. A reduction in delay (negative delay) is needed to offset the BBC HD sound delay problems.

My understanding is that AV should automatically be in sync when a tuner is connected to a TV via an HDMI connection. No adjustment for sound delay should be needed.

I note you say that travelling bars cannot be used to measure sync as interlace and flat screen de-interlacing processes make the result unpredictable and therefore invalid. However, please note that my "sync by eye" check using the travelling bar does give the same result as the photodiode measurement. (Note: the photodiode measurement can only be accurate to + or - 5ms as the TV’s backlight is 100Hz with a 50% duty cycle.)

Thank you for confirming that the BBC’s test signal is calibrated to be 0ms measured on a CRT using the top bar, or 1ms audio early if measured on the second bar. I realise that my TV is an LCD; not a CRT, but I would still expect the received test signal to be 0ms at the top bar; not the 70ms sound delay I measured.

Over the weekend I did not notice any obvious delayed sound on the BBC HD channel programmes I watched. So I’m hoping to catch the test signal again to see if the results are now different.

In the meantime I ask these further questions as the answers might help unravel what’s going on.

Please can you confirm what is the top bar. Is it a single line in 1 field or 2 lines in 1 frame? If 2 lines, does the sound follow 0ms after the first top bar line is lit? Or, again if 2 lines, is it 0ms after both top bar lines are lit (i.e. 20ms after the 1st line is lit)? [Note: My photodiode picks up light from more than one line. It covers about 3 display lines on my TV]

Also, given the position of the travelling bar in the screen, please can you confirm whether the scale has been positioned to allow for this area of the screen being written later than the top bar (about 15ms later for an interlaced display). Please also confirm whether the travelling bar is moved every field (i.e. every 20ms) so that 100Hz (or 200Hz) TVs can properly render the extra (intermediate) frames they show.
A Christmas Present from the HD Channel! 1:50pm on 30 Jan 2009 As requested here is my BBC HD Sync Test Result

Date: 30/01/09
Time: 11:50 AM
Receiver Setup: Humax Foxsat HD -> HDMI -> Philips 47PFL9632D
Firmware: TV 64.14.18, Foxsat 1.00.15
HDMI: TV has version 1.3a (i.e. it supports the auto lip-sync feature).

Test Method.
A stereo .wav file was recorded with the TV audio occupying the left channel and the amplified photodiode output occupying the right channel. The resulting file was analysed in the time domain using Nero Wave Editor.

Test Results.

Sync Flash Bars.
The top bar was visible. The sound starts 70ms after the top bar is lit.

Measurement Bar.
The photodiode was placed just above the lower calibration marks in the path of the measurement bar. At the zero point the sound commenced 40ms after the measurement bar passed the photodiode. At the sound late 1 frame point the sound commenced at the same time as the measurement bar passed the photodiode.

Comments.
0. The results accord with my perceptions on viewing the BBC HD sync test.
1. The BBC HD channel is the only one on which I notice that the sound is delayed. ITV HD and all SD Freesat channels do not have a noticeable sound delay. I notice no sound delay on Freeview channels (TV internal tuner, no HDMI).
2. This TV has HD natural motion for extreme motion sharpness. In it displays a new frame every 10ms.
3. Both Humax and Philips tell me that the BBC HD sound delay problem is not due to a fault in their products.

Questions.
Why is the sound delay greater when measured against the sync flash bar at the top of the picture? Do the BBC synchronise their sound to start after a complete 2 field 40ms frame has been displayed? If so, given that the BBC HD broadcast is 1048i, would it be better for the BBC HD sound to start after just one 20ms field has been displayed?

Request.
Please would the BBC work with the receiver and TV manufacturers to eliminate the long running BBC HD sound delay problem first reported by me in October.

«Newer
Older»

^ Back to top

BBC MMVI

The BBC is not responsible for the content of external internet sites

BBC.co.uk

BBC Blog Network

Find and talk to the BBC's bloggers.

User Profile

Viewer00