SpinVox uses humans for voice-to-text conversion

The voice-to-text firm Spinvox uses people, as well as automated systems, to transcribe messages in call centres in South Africa and the Philippines, the BBC has reported.

The report, published on Thursday, also suggested that the Buckinghamshire-based firm is under financial strain as a result of using the call centres in question. The BBC said it had received its information from a source inside SpinVox, as well as from a former employee of one of the call centres.

SpinVox sells software that lets users read received voice messages in text form. The company opened up the application programming interfaces (APIs) for its platform in February, in a bid to get other companies to build SpinVox's functionality into their own applications.

On its website, SpinVox said its conversions are processed by a system called D2, or 'The Brain'. "D2's pretty smart," the website reads. "It's bound to be, as D2's a combination of artificial intelligence, voice recognition and natural linguistics. But it also knows what it doesn't know and is able to call on human experts for assistance."

The Information Commissioner's Office (ICO) told ZDNet UK in a statement that it would contact SpinVox over the data-protection implications of the BBC's story, "especially with regards to the transfer of personal data outside the European Economic Area".

"The Data Protection Act does not prevent SpinVox from using people, rather than machines, to translate audio messages into text," the ICO's statement reads. "However, it may be helpful if the company is clearer about the likelihood that people will be used to translate messages. This is particularly important if customers are using the service for transmitting sensitive or secure information."

SpinVox did not respond to several requests from ZDNet UK for comment on the report.