speech recognition system

[munyandi]

hi every one!!!

am doing a project that convert spoken word to text ( speech recogniton) in java... am a semi - beginner in programming. i wuld appreciate any type of help from any body. pls help me out???

[Ezzaral]

Learn a lot more programming and signal processing, as speech recognition is far from trivial to implement.

[jwenting]

maybe you should first attempt a project focussed at writing grammatically correct English before trying a project at interpreting the spoken word.

I can confidently say, you will not be able to do this.

But please, go ahead and prove me wrong.

[Colin Mac]

Depends on whether you want to use the mic insert on the PC or external hardware and send the voice info to the PC.
The latter is probably doable for word recognition but not translating whole sentences.

[sciwizeh]

first be thorough in learning java. then check out the Java Speech API

[munyandi]

yeah am planning to use the mic as an input method.

ok guys stopn discouraging me now, instead show your support.. so far my program can detect the availability of the mic and return the number of mixer installed.

so guys can u pls help me with the recognition part...
i have read the java speech API but it wasn`t too helpful....

[Ezzaral]

I already offered advice on that - learn audio signal processing and perhaps some machine learning theory.

No one here is going to be able to teach you speech recognition code. It's a very complex topic that you are going to need to research for yourself and experiment with.

[mahlerfive]

The reason people are discouraging you is because the very basics of this topic might be material for a senior undergrad course or a graduate course, and considering you said you were a semi-beginner programmer, you probably aren't quite ready for something of this magnitude.

Before even starting to code anything you will have to learn quite a lot of theory. As Ezzaral said, you need to study audio signal processing to begin with. If your goal is to recognize single words through training, then you will probably need to learn about neural networks as well.

I would say that if you are trying to recognize speech in sentences - trying to do this with a decent accuracy would make for a good graduate thesis... if it was a fairly trivial problem, we would already have near perfect speech recognition, but instead the best we have is some fairly clunky speech recognition that only works somewhat well when it is trained a lot, and even then it will only work with that one person and with little to no background noise.

[MaxDev]

First Check out this site:
http://msdn.microsoft.com/en-us/library/ms713505.aspx
Learn Windows Multimedia API.

[scorpious]

Hellow, I think we have the same problem, I am planning to develop a speech to text pure Filipino- Filipino dictionary, it is for my fourth year research and development project please send me some advise at [e-mail],