Recognize only one sentence

[Juan Albéniz]

I need speech recognition technology if I only want that the PC compare a
sentence?

The PC has the "Hello world" sentence pre-recorded by "speaker X".

After that a "speaker A" says "Hello world", then the PC has to say right or
wrong.

Can someone help me? website, document, something...

Or I have to use any algorithm with the F0, formants, intensity, etc etc...

Thank you.

[Nickolay V. Shmyrev]

I need speech recognition technology if I only want that the PC compare a
sentence? The PC has the "Hello world" sentence pre-recorded by "speaker X".
After that a "speaker A" says "Hello world", then the PC has to say right or
wrong. Can someone help me? website, document, something... Or I have to use
any algorithm with the F0, formants, intensity, etc etc... Thank you.

This thing is done by Dynamic Time Warping algorithm. There is no support for
DTW in Sphinx4.

http://en.wikipedia.org/wiki/Dynamic_time_warping

[Juan Albéniz]

Ok I know about DTW, but I need to compare a voices by different speakers, so
the DTW only works to align the waves.

When I have aligned the two waves, I have to compare what? because I'll have
an array pull with the characteristics of the wave any time.

Which characteristics I have to extract from the wave... low frequencies,
intensity... this is my problem, I don't know which are the best
characteristics to compare two waves spoken by two different speakers. Thanks
for help me nshmyrev.

[Nickolay V. Shmyrev]

I don't know which are the best characteristics to compare two waves spoken
by two different speakers.

Usually applications compare MFCC features.