learn app to respond to specific voice

[brax]

Hello Halle,

I’m not a developer my self but are currently looking for a developer that’s able to help me creating an app with voice recognition capabilities. I know you’re not for hiring unfortunately, but as I don’t know the limits of OpenEars I hope you could answer my question so I know what I could ask of a developer.

Is it possible with OpenEars to create an app that requires the user to speak the words that should be recognized first and assign a specific user profile to the voice of that user so the app only recognizes when that specific user speaks the words again and therewith ignores any other voice from other users as long the profile of that specific user is active?

Thanks you for your reply!

[Halle Winkler]

Hello brax,

Sorry, user voice identification isn’t part of the feature set of OpenEars. What you would want to focus your research on is called ‘speaker identification’ and ‘acoustic fingerprinting’, and probably the best place to start searching for it, in order to get a sense of the state of the art, would be to search for those keywords on Stack Overflow at http://www.stackoverflow.com.

[brax]

To bad OpenEars can’t be used for this functionality. I really appreciate your response and tips what to look for! Thanks!

[Halle Winkler]

You’re welcome, good luck!

[iced]

Hey Halle,

The topic seemed relevant to my question and I decided to not open a new topic and ask it here.

Is it possible to calibrate OpenEars to recognize some harder-to-recognize English accents?

[iced]

Btw I don’t mean a specific thing with “harder-to-recognize accent”. My question is if it is possible to perform User-calibration. Thanks

[Halle Winkler]

It isn’t possible to train to a specific user, but you can adapt the entire acoustic model. So, I think the answer is that no, you can’t do user calibration, but it would actually be possible to adapt to a harder-to-recognize accent.

[Author]

Posts