Compare 2 Recorded Voices?

Question

Hello everyone, Good morning.

I am actually working on a app (for iOS and android) that will teach you the "Tajweed" (accent) of alphabets (In Arabic) and how to speak those alphabets.

So, what i need here? it's really a complicated thing but let me explain it to you. firstly i will record a voice from actual correct accent (It would be just a alphabet of arabic like "Alif", it will not be the whole song or sentence here) and save it into the app, now let user to speak, app gonna record his voice. and compare the user voice from actual voice.

Here are the parameter i need to compare.

Length of the speech (not length of the audio, means app should pick up just a speech).
Pitch of the speech.
Amplitude of the speech.
Anything that can compare the alphabet, like he spoke "alif" and any other parameter that will be useful in this comparison, as you know what i actually need.

here is the example app for same kind of stuff: https://itunes.apple.com/us/app/american-accent!/id520326878?mt=8

Thanks in advance, and it will be appreciable if anyone could help with the code.

Answer 1

Answer by tanoshimi · Jan 27, 2015 at 11:00 AM

I don't see what this has to do with Unity - it's a general programming algorithms question. You might want to read about how Shazam approach their music recognition service - by creating a fingerprint of each song in their database and then comparing the spectrogram of a given input to find potential matches - you could try a similar approach with your spoken samples: http://www.ee.columbia.edu/~dpwe/papers/Wang03-shazam.pdf

Add comment · Share

Compare 2 Recorded Voices?

1 Reply

Follow this Question

Related Questions