Speech recognition for attendance

I want to develop a matlab code for attendance based on speech recognition. So I want to do speech recognition using matlab.

What should I do? and where should I start?

Just tell me the procedure. I will do the rest.

[Walter Roberson]

Based upon your question after this one, it appears that you want to work with "voice recognition" rather than "speech recognition".

Answer 1

[Kaustubha Govind]

If you are willing to use existing libraries and are developing on Windows, you should be able to take advantage of the Microsoft Speech API. There are some submissions on File Exchange that show examples of how you can call into the API: http://www.mathworks.com/matlabcentral/fileexchange/?term=microsoft+speech+api

Answer 2

[i Venky]

I know the text to speech conversion of Microsoft. Thanks for telling me that. But I also want to convert speech to text. In addition I want to recognize the speech too. How would you do that? Where should I start?

[Walter Roberson]

Do you want to do _voice recognition_ or _speech recognition_ ?

I don't know how you plan to handle,

"Merry Mary, marry!"

[Sean de Wolski]

Or their they're there

[Kaustubha Govind]

iVenky: I believe the Microsoft API does perform speech recognition as well. The link in my answer takes your directly to the speech recognition part of the API:

"The Windows Desktop Speech Technology software offers a basic speech recognition infrastructure that digitizes acoustical signals, and recovers words and speech elements from audio input."

[i Venky]

Acutally I made a mistake while asking the question. It's speaker recognition and not speech recognition. Anyway thanks for your response.

[i Venky]

I am going to use MFCC and GMM for speaker recognition. Which method is best for speaker recognition?

Answer 3

[Walter Roberson]

If I understand correctly, you wish to take attendance using voice recognition.

I wonder, though, how you plan to deal with colds or laryngitis?

I also wonder what age range this would be targeted at? If this would be used for teenagers, then voices can change substantially even from day to day.

[i Venky]

In the beginning I am just going to try speaker recognition. If possible I will try attendance using that. Yes, you are right, there will be more practical difficulties.

Thank you for your suggestions Walter Robertson.