LinuxQuestions.org
Did you know LQ has a Linux Hardware Compatibility List?
Go Back   LinuxQuestions.org > Forums > Non-*NIX Forums > Programming
User Name
Password
Programming This forum is for all programming questions.
The question does not have to be directly related to Linux and any language is fair game.

Notices



Reply
 
Search this Thread
Old 07-21-2011, 05:14 PM   #1
ridoy
LQ Newbie
 
Registered: Feb 2011
Posts: 19

Rep: Reputation: 1
Unhappy voice recognition


I'm doing a project of voice recognition.I want to spell a word and the computer will write it to a simple notepad application of my project.During coding what steps i should follow that my computer will able to write my spelled words?Though i know i have to build up a grammer written in xml for this.I'm doing my project in C sharp and i use windows 7.Anyone can advice properly..?

Thanks for any help.
 
Old 07-22-2011, 03:45 AM   #2
Thor_2.0
Senior Member
 
Registered: Nov 2007
Location: Somewhere on my hard drive...
Distribution: Manjaro
Posts: 2,227
Blog Entries: 23

Rep: Reputation: 279Reputation: 279Reputation: 279
(wisely edited out)

Last edited by Thor_2.0; 07-22-2011 at 03:54 AM.
 
Old 07-22-2011, 04:33 AM   #3
gnashley
Amigo developer
 
Registered: Dec 2003
Location: Germany
Distribution: Slackware
Posts: 4,775

Rep: Reputation: 481Reputation: 481Reputation: 481Reputation: 481Reputation: 481
What you mean is 'speech recognition' or STT (speech-to-text). It happens to be one of the most difficult things to do with a machine.

There a couple of open-source projects which already do what you propose. The best one is probably 'julius'. And then there is 'espeak' and 'sphinx'. There is a fairly-developed voice-command system called 'simon' which uses on of the above.

The other side of speech applications is called TTS (text-to-speech). The best and largest and most-complex is 'festival, but there is a lite version called 'flite'.
'simon' uses festival. IIRC, espeak has both SST and TTS.

I use julius for speech recognition as it seems to be more accurate than sphinx. I then process the text created using a small AI program, and then use flite to output speech from the output of the AI processing.

I'm not sure which of any of these programs can be compiled or is available for Windows. Even if you have to write your own program, referencing the sources for the above programs will help you a lot. In fact you probably will wind up using much code or data files from other people. You mentioned a grammar, you need a 'language model' which gets created using a grammar for the language you want to interpret, and then you need a bunch of recordings of examples of people speaking the words contained in a prepared lexicon. The the recordings get analyzed so that graphs can be created which 'show' the program what each letter or word sounds like. The result is the 'speech model'. The best speech models are the ones which use the data of speech from as many people as possible. Or you can buy prepared speech models. Most open-source projects use speech data from voxforge.
 
Old 07-24-2011, 12:13 PM   #4
ITPhoenix
Member
 
Registered: Mar 2011
Location: Northeast USA
Distribution: Debian Squeeze
Posts: 47

Rep: Reputation: 1
Cool

Quote:
Originally Posted by ridoy View Post
I'm doing a project of voice recognition.I want to spell a word and the computer will write it to a simple notepad application of my project.During coding what steps i should follow that my computer will able to write my spelled words?Though i know i have to build up a grammer written in xml for this.I'm doing my project in C sharp and i use windows 7.Anyone can advice properly..?

Thanks for any help.
You would have better luck speaking the word naturally and having it printed to text. The engines are poor with beginning utterances to begin with. Then you would have to write a program that assembles the letters into a word.
Having looked into this, it appears Sphinx4 is the best bet for you. It is written entirely in Java and requires JDK and Ant as prerequisites. So you do all your application programming in Java as well, which will "simplify" things. This program also runs better on Linux, and there is minimal support to begin with and even less for Windows.
Before all that, you would have to be fairly well versed in how the recognizer and general language processing works. Then on to Java....

I'm having trouble even installing the thing....
 
Old 07-25-2011, 04:42 PM   #5
ridoy
LQ Newbie
 
Registered: Feb 2011
Posts: 19

Original Poster
Rep: Reputation: 1
Thanks all of you for your comments.I have done this with C sharp and Windows Speech Recognition machine.
 
Old 05-15-2012, 12:45 PM   #6
fakie_flip
Senior Member
 
Registered: Feb 2005
Location: san antonio, texas
Distribution: Fedora 64 bit RAID0 + LUKS, CentOS (server), Backtrack, Gentoo Hardened
Posts: 1,441

Rep: Reputation: 81
I have done this in Java.

http://linuxinnovations.blogspot.com...to-speach.html
 
Old 05-16-2012, 01:05 AM   #7
ITPhoenix
Member
 
Registered: Mar 2011
Location: Northeast USA
Distribution: Debian Squeeze
Posts: 47

Rep: Reputation: 1
Java TTS; interesting. This would make it easier to incorporate in any Java program. I wonder if arctic voices could be added. I love the slt voice. Sounds like an educated, 25 yo female from northeast US.
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
voice recognition georgekjolly Programming 2 10-25-2005 03:35 PM
There MUST BE voice recognition! CatBreath Linux - Software 5 03-26-2005 12:16 PM
voice recognition pantera Programming 0 10-20-2004 07:17 AM
voice recognition pantera Programming 2 07-03-2004 06:21 AM
voice recognition ferreter Linux - General 0 03-17-2003 07:22 PM


All times are GMT -5. The time now is 12:46 AM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
identi.ca: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration