Well since last month I been working on a project to do a "Speech Recognition System" without the use of any 3rd party help like Speech API.

So, this is what I have done in my project. Speech recording, save it to a wave file. Reduce the noise of the file. And then initially I was gonna compare and match the wave file by its Frequency. After a few days of trying , I found out that there might be some issues about it.

So here's the question,
1. what other way could I finish my project?
2. Is comparing the frequency of the wave files still possible?
3. Is it possible if I JUST add up the comparing part of the Speech API? ( I would definitely try to avoid this if possible but just as an backup plan. )

http://en.wikipedia.org/wiki/Digital_signal_processing
But if you're trying to do this yourself, then get comfortable for some serious maths.

> 2. Is comparing the frequency of the wave files still possible?
I've no idea.
But the simpler your answer is, the more likely it is that it will be very specific to only one voice at once (and even then, it might take a lot of training runs).

>3. Is it possible if I JUST add up the comparing part of the Speech API?

Yes but don't say how? In fact I suggest you to use SAPI.

if I were to you the SAPI , which part should I focus on , I don't hope to BIN all my efforts and redo everything again with SAPI.

Will be focusing on comparing two WAVE FILES. Very simple words like.

APPLE
BALL
CHICKEN

This article has been dead for over six months. Start a new discussion instead.