Changes between Version 13 and Version 14 of AcousticModels


Ignore:
Timestamp:
01/11/07 13:11:23 (16 years ago)
Author:
kmaclean
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • AcousticModels

    v13 v14  
    1 == Free Acoustic Models == 
     1= Free Acoustic Models = 
    22  * [http://htk.eng.cam.ac.uk/develop/atk.shtml ATK distribution] contains compiled HMM models! - not sure if the compiled version is usable with Julius. 
    3   * Keith Vertanen's free Acoustic Models (no source audio) using Wall Street Journal (WSJ) Corpus: 
    4     * [http://www.inference.phy.cam.ac.uk/kv227/htk/acoustic_models.html Trained acoustic models for HTK] (in HTK binary - Julius cannot read HTK binary format) 
    5     * [http://www.inference.phy.cam.ac.uk/kv227/htk/acoustic_models.html Trained acoustic models for Sphinx] 
     3 
     4 
     5== Sphinx == 
     6  * English 
     7    * Sphinx distributions include English Acoustic Models 
     8    * Keith Vertanen's free Acoustic Models (no source audio) using Wall Street Journal (WSJ) Corpus: 
     9      * [http://www.inference.phy.cam.ac.uk/kv227/htk/acoustic_models.html Trained acoustic models for Sphinx] 
     10  * French 
     11    * [http://sourceforge.net/project/showfiles.php?group_id=1904&package_id=208123 Sphinx French Acoustic and Language Models] 
     12 
     13== HTK/Julius == 
     14 
     15    * Keith Vertanen's free Acoustic Models (no source audio) using Wall Street Journal (WSJ) Corpus: 
     16      * [http://www.inference.phy.cam.ac.uk/kv227/htk/acoustic_models.html Trained acoustic models for HTK] (in HTK binary - can be converted to ASCII so Julius can read them - see this [http://www.voxforge.org/home/forums/message-boards/general-discussion/can-we-use-th-hub4-or-wsj-thats-supported-by-sphinx link])