Rory Cellan-Jones

Spinvox: A question of Nuance?

  • Rory Cellan-Jones
  • 14 Dec 09, 14:10 GMT

Remember Spinvox? The British voice-to-text start-up has been out of the news since the summer, when we revealed some serious questions about its technology and its finances.

Screenshot of Spinvox websiteNow the Sunday Times has reported that the company is close to a £92m sale to the American voice technology giant Nuance.

A source at Spinvox told me a few weeks back that talks were under way with Nuance - but there had been rumours of a sale to just about every big technology name you could mention. This time, though, it looks as though a deal really is in the offing.

For one thing, the Spinvox PR machine is not rubbishing the story; for another, today's deadline for the company to repay a £30m loan has been put back to the end of January, apparently to let the negotiations proceed.

The real questions are what exactly Nuance is considering buying - and why. Back in the summer, on the very day my first article about Spinvox appeared, Nuance got in touch with me.

They wanted to boast about their own world-leading technology, while implicitly criticising Spinvox, both for its claims about its ground-breaking speech recognition work and for its attitude to privacy.

Nuance told me in July that Spinvox was "absolutely not ahead of the game", and the company's spokesman was also keen to stress that "in the current climate about data privacy, being transparent is absolutely critical."

So why, just a few months later, would Nuance be interested in buying the Spinvox?

Well, there are relationships with some of the world's biggest mobile operators - notably a giant contract to supply a service to up to 100 million subscribers in South America.

Those kind of deals take years to hatch, so Nuance may see Spinvox as a short cut to a lucrative relationship with the likes of Vodafone and Telefonica.

Then there's Spinvox's Voice Message Conversion System, D2 - or "The Brain", as the company dubs it.

Speech experts at the firm still insist they have developed something innovative, even if it's more a case of humans aided by technology than technology aided by humans.

But the real value may lie in all the data that Spinvox has accumulated over the years, in the form of millions of voice recordings and the text derived from them.

How much of it is still stored on the company's servers is unclear - although I've just noticed this line in the Spinvox terms and conditions:

"We may establish general practices (and change such practices without notice) and limits concerning use of the Services, including, without limitation, the maximum number of days that messages or other uploaded Content will be retained by us."

It also makes clear that this data may be sent around the world:

"We may transfer your information outside the European Economic Area. Again, we will endeavour to comply with the Data Protection Act in respect of such transfers."

What I can't find is any reference to what happens to your old messages if Spinvox is sold to another business.

So if this deal does go through, customers will want to know what it means for the privacy and security of their data.

Shareholders will also be asking what a sale to Nuance would mean for them. Some early investors - such as Carphone Warehouse and Invesco - have already effectively written off most - if not all - of their money.

And what of those Spinvox staff who heeded appeals this summer from their boss Christina Domecq to take part of their wages in the form of stock options to help the business through to profitability?

A number of those employees have now left, but they may now be pulling those share options out of a drawer and wondering whether they are worth anything.


  • Comment number 1.

    Nuance are in the desktop and vertical application arena one of the world leaders, especially with their price points and a few of their vertical markets.

    Clearly if they are this far down the road they see some advantages to the way Spinvox works or the contracts and relationships they hold.

    Does that goodwill equate to £90m plus? Probably not, there has to be more under the hood to make the investment worth it.

    Many people find voice recognition an advantage, it saves a secretary, or their fingers from typing, but it's never hit the masses as we've always been told it should.

    I'm in a group of people, not that small a minority, one in seven of the population who have a significant hearing loss. I am profoundly deafened in both ears.

    Over the past couple of years SpinVox has enabled me to communicate with people when their initial communication was to phone me. I can't use the telephone and depend on this technology.

    I use Dragon from Nuance and have tried to utilise it during meetings to provide a rough realtime translation of a meeting - without teaching it the six different voices in the meeting, making people slow their speech right down so that this software can work with it and then only handle one voice profile at a time is the weakness.

    Take SpinVox and my voicemails, people with accents and dialects from all over the UK and the world can have their voicemail delivered to me as a text or an email with the key data encoded. There is a huge learning curve for any software packager to do this work, try using Nuance for yourself and really test it, you will spend hours correcting and educating it for your voice in a controlled environment, not at the end of a poor mobile line.

    You raise questions of security, the vast majority of voicemails are one sided, a simple answer to a previous question or the question needing an answer. Other than that it's 'Hi can you call me when you get this'.

    There are some high flyers who have a few confidential voicemails, they are also likely to be the people who never change the default voicemail passwords on their mobile account either.

    Hopefully the combination of services will see a platform be delivered that will offer more to a wider spectrum of users, for more applications than just a few key verticals that we have today.

    Voice recognition has been around for a long time and the few players here today have made little real progress with it's real development, newer and faster processors have delivered faster processing and the opportunity for more error checking, closer to real time performance than before.

    I for one look forward to the combination of these two companies and their technologies.

  • Comment number 2.

    For their £90m they will get the contracts, customers and most importantly get rid of a potential competitor.

    Can't see much milage in the data myself, there will be very little there of interest to anyone and would take years to sift through.

  • Comment number 3.

    I find it hard to believe Nuance would be that stupid…..

  • Comment number 4.

    Me too, 90M is an awful lot of money to spend on these guys. Surely they can drive a better deal than that! I can't imagine there are companies lining up to buy SpinVox given the now obvious flaws with their business model so why on earth they would spend this sort of amount is beyond me. What are the existing investors going to do if they dont strike a deal? Pump another 100M in, dont think so. As a shareholder in Nuance I sure hope their M&A guys are smarter than that, they have Spinvox over a barrel so if they want them let's see a sensible price paid and £90M given that this is effectively a fire sale looks way too dear to me.


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