Looking for info about sound to text conversion

[ykishore]

speach to text romanian

Hai! I'm kishore from India.
I have been hearing of the sound to text direct conversion from lot many days. So, if anybody knew information about it please reply me.

urgent :? :?

 

[DarkJedi]

Re: Sound to text

Well, to me it sounds to far in the future, if i were you start looking for speech recognition, it'll give you some sort of idea on how it's done, lot's of DSP concepots are involved on recognition and to convert to text you need to recognize every consonant and vocal, i really don't think that is done, at least not like a product you can buy now.

 

[mami_hacky]

Sound to text

You can study some good books on this topic, among them, "Digital Speech processing, Synthesis and Recognition" I think, By Furui. Another valuable source by Deller, I think is "Digital Speech processing" IEEE Press.
I'm wondering if you are asking for a speech recognition system. what do you mean by sound to text conversion? Why you have not written Speech to text conversion?
There are some systems capable of showing the manner of a person as words. Analysing how a person is speaking you can understand if he/she is sad or happy, if he/she is crying and .... these kind of systems are suitable for applications like an automatic system for taking care of a baby .... Using Wavelets beautiful tasks like what I said can be done.
Cosider HMM, Hidden Markove Models are the base of almost all of speech recognition systems. HTK, is a toolkit for acting with HMMs. it is a famous tool. it has a site for itself. have a look at it. studying the HTK documentation you will certainly understand how a good speech recognition system works. you will understand the role of grammar, words, ....
I think it is enough for now, however there is much much more to say.
Take care, if you are going to build a speech recognizer for your own language then the hardest problem to solve is to obtain a phonetically balanced dictionary of spoken words.

 

[ykishore]

Re: Sound to text

My idea pertaining to this is voice to text conversion only. Sorry for posting it as sound to text conversion.

Moreover you have mentioned that there is a web site that might guide me in this process, but, you haven'tmentioned the address of the web.

I would feel happy if you do that.

Thank you for posting reply to my question.

 

[zkai2000]

Re: Sound to text

this sounds interesting. it'll be a great technology for the deafs!
do u have any info on this topic? mind posting them to share?
sound and speeches are actually waves and PCs take in the signals...issit possible to convert them into characters showing the words being pronounced?

 

[eltonjohn]

Re: Sound to text

One thing very important before doing the real recognition of the speech is tho acquire the sound .A great deal of effort goes also into this .The signal has to be filtered an clean of ambient NOISE .. is not just a MIKE and a crummy amplifier , But a sofisticated signal processing of the input Before is feed to the real recognition algorithms .
So i suggest to start dealing with this .before anything else .otherwise it won't matter if you get the guillespie DSP recognition engines .It WON'T PERFORM a its best!
a useful link :
**broken link removed**

 

[1stfaruk]

Re: Sound to text

i am aslo interested in this project , and it requires high voice quality and ypu should have some information about the voice signal processing.

 

[khubaibahmed]

Re: Sound to text

Look for STT (speach-to-text) algo on net or TTS (Text-to-Speach) algo