Babbel introduces speech recognition to aid language learning

Babbel, the language learning site, has added “realtime” speech recognition to enhance its practical application and enable users to fine-tune their pronunciation skills. This pits the service up against more traditional players such as TellMeMore or Rosetta Stone, says the company.

The speech recognition functionality was built in-house – much of the team’s background is in audio technology – although it was realised with the latest 10.1 update to Adobe’s Flash plug-in, which enables developers to access audio data captured from the user on the client-side instead of streaming to a back-end server for analysis. For realtime feedback, local processing is preferable, says Babbel, and had Adobe not offered this option, the company would need to have built its own browser plug-in, which is hardly ideal.

Included in all Babbel courses, the browser-based speech analysis is designed to give learners “instant evaluation” of how close their pronunciation is to that of a native speaker and to tackle the problem that many students of foreign languages have in that they lack speaking practice and the confidence that comes with it.

In the associated exercises, learners listen to a word or phrase and are then prompted to repeat it back. They are then given a rating of 0-100, with a result of 50 or higher meaning that they are “generally understandable”, after which they can move on to the next exercise or choose to have another go.

Moving forward, Babbel plans on bringing its newly developed speech recognition functionality to its iPhone apps.

The Berlin-based company, which was founded in August 07, says it has 700,000 registered users from over 200 countries, while its iPhone apps have passed the 100,000 download mark. It doesn’t, of course, disclose how many of those users are paying customers but does claim to be profitable since dumping its freemium model in November 09.

Babbel (via Lesson Nine) is backed by Kizoo AG and VC-Fonds Berlin.