Skip to content

munggok/Indonesian-Speech-Recognition

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

README

I. DESCRIPTION

This is a repository for my Speech Recognition class using HTK and Festival for Indonesian language.

II. SPEECH RECOGNITION

I used HTK for speech recognition task that can be downloaded from http://htk.eng.cam.ac.uk/. You may need the manual (HTK Book) and you also may want to start from Chapter 3 Tutorial if you are only interested in the experiment without knowing the basic theory (not recommended).

My training data (train.tar.gz) contains 50 recorded sentences together with their transcriptions. The test data contains 20 sentences (or 25?). Each folder of train and test data has some slightly different configuration. Read Notes (Directories) for the details.

The best results is 90% (18 out of 20 sentences) after some tweaking with major help from my professor.

III. SPEECH SYNTHESIS

Speech synthesis was performed using Festival and Mbrola voice database, since Festival does not provide Indonesian voice. The Indonesian voice database in Mbrola was created by Arry Akhmad Arman (http://indotts.melsa.net.id/).

The speech-synthesis folder only contains one important file 'id1_mbrola.scm'. This is a festival wrapper to use Mbrola voice database inside Festival. I think this is the first one, because I cannot find any other scm file for Indonesian.

The result is still poor, mainly due to intonation and accent rule, which is none.

IV. DISCLAIMER

This is a school project, do not use it for any important task. I do not responsible for the undesired results of using the data provided here.

About

Repository for my Speech Recognition class

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Scheme 100.0%