Monday, May 21, 2007

A/E Speech DataBase

Arabic/English Speech Database

This is the first step to build Speech Recognition system , We should create a database for training and testing any algorithmes you used.

English Speech Database

After a lot of searches on the internet , I found that most of speech databases are not free and you should buy it , this site: http://www.ldc.upenn.edu/ contain many speech databases for example TIMIT database.

But Finally I found one Free here Speech separation challenge

http://www.dcs.shef.ac.uk/~martin/SpeechSeparationChallenge.htm

The training and development sets are drawn from a closed set of 34 talkers of both genders. this document describes the corpus in more detail
training data from individual talkers: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

The last challenge ,We need to segment the speech wav files to words , I do it manually using very efficient program "Adobe Audition 2.0 ".

Arabic Speech Database

it is a big problem to create your own database , if you work alone and your friends doesn't want to help you and Disheartening.

I'm thinking to new method to build database without need anybody , ISA I will explain the new method after I make sure that it work properly.

**you can build your own Arabic Speech database from the Holy Quran recording.

No comments: