Changes between Version 6 and Version 7 of HMMVoiceCreation


Ignore:
Timestamp:
04/29/08 17:13:16 (17 years ago)
Author:
marcela_charfuelan
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • HMMVoiceCreation

    v6 v7  
    11 
    2 ''' 
    32 
    4 == Creating HMM voices for MARY with the HMM Voice building tools. == 
    53 
    6 ''' 
     4= '''Voice Import Tools Tutorial : How to build a HMM-based voice for the MARY platform''' = 
    75 
    8 The steps for building a HMM voice for Mary system can be summarised in:[[BR]] 
     6For creating HMM-based voices we use a version of the speaker dependent training scripts provided by [http://hts.sp.nitech.ac.jp/ HTS] that was adapted to the MARY 
     7platform. The steps for building a HMM voice for the MARY platform can be summarised in:[[BR]] 
     8 
    99I)   Checking the necessary programs and files[[BR]] 
    1010II)  Data preparation[[BR]] 
    1111III) Training of HMM models[[BR]] 
    12 IV) Adding a new HMM voice in the Mary system.[[BR]] 
     12IV)  Adding a new HMM voice in the Mary system.[[BR]] 
     13V)   Creating other voice in German or English (__if you want to train HMMs with another speech database__).[[BR]] 
    1314 
    14 The following is an explanation of these steps for running the HTS-demo_CMU-ARCTIC-SLT (http://hts.sp.nitech.ac.jp/archives/2.0.1/HTS-demo_CMU-ARCTIC-SLT.tar.bz2) 
    15 using the Mary system. 
     15The previous steps will be explained creating the HTS cmu-arctic-slt, speaker independent, demo voice: 
     16http://hts.sp.nitech.ac.jp/archives/2.0.1/HTS-demo_CMU-ARCTIC-SLT.tar.bz2 
    1617 
    1718The training scripts used here are the latest versions, that is, it is required HTS_2.0.1 and SPTK-3.1. Some scripts have been added-modified to:[[BR]] 
    18 - Use Mary instead of festival as text analyzer.[[BR]] 
     19- Use MARY instead of festival as text analyzer.[[BR]] 
    1920- Train bandpass voicing strengths for mixed excitation.[[BR]] 
    2021- Process language specific settings as parameters.[[BR]] 
    … …  
    2829- MARY TTS Recent Version - Download Link: http://mary.dfki.de/Download [[BR]] 
    2930- Openmary - SVN from http://mary.opendfki.de [[BR]] 
    30 - Mary patch for HTS demo: HTS-2.0.1-demo_CMU-ARCTIC-SLT_for_Mary-3.5.0.patch [[BR]] 
     31- MARY patch for HTS demo: HTS-2.0.1-demo_CMU-ARCTIC-SLT_for_Mary-3.5.0.patch [[BR]] 
    3132 
    3233 
    … …  
    4748Here it is used: HTS-demo_CMU-ARCTIC-SLT.tar.bz2 for HTS-2.0.1 
    4849 
    49 0.2) download and unzip the patch file for using Mary instead of Festival text analyser. 
     500.2) download and unzip the patch file for using MARY instead of Festival as text analyser. 
    5051 
    5152https://mary.opendfki.de/repos/trunk/lib/hts/HTS-2.0.1-demo_CMU-ARCTIC-SLT_for_Mary-3.5.0.patch.zip  [[BR]] 
    … …  
    107108the same as you would do when running "make configure" with the original HTS scripts. 
    108109 
    109 8- Run HMMVoiceMakeData component of the HMM Voice trainer group to run the HTS procedure "make data". This procedure is the same as in the original scripts with additional sections for calculating strenghts (for mixed exitation), global variance, and handling of Mary context features. 
     1108- Run HMMVoiceMakeData component of the HMM Voice trainer group to run the HTS procedure "make data". This procedure is the same as in the original scripts with additional sections for calculating strenghts (for mixed exitation), global variance, and handling of MARY context features. 
    110111 
    111112Particular procedures can be repeated isolated, fixing the particular settings for this component. For example, if the procedure that creates strengths (str directory) has to be repeated with a different set of filters (data/filters/), please set: [[BR]] 
    … …  
    116117The procedures can be repeated manually as well, going to the data directory and running "make data" or "make str", as is normally done with the original HTS scripts. 
    117118 
    118 Note: the Makefile in data/ includes a gv: section copied from HTS-2.1alpha version to calculate global variance files. In Mary, this files are generated little endian and contain a header of size one short to indicate the size of the vectors it contains. 
     119Note: the Makefile in data/ includes a gv: section copied from HTS-2.1alpha version to calculate global variance files. In MARY, this files are generated little endian and contain a header of size one short to indicate the size of the vectors it contains. 
    119120 
    1201219- Run HMMVoiceMakeVoice component of the HMM Voice trainer group, here again particular training steps can be repeated selecting them (setting in 1, all the others in 0) from the settings of this component. This is equivalent to run again:  [[BR]] 
    … …  
    168169- Rename this directory as your new voice name, for example german_voice, and delete the directories data/raw and data/utt. 
    169170 
    170 - Apply the Mary patch to the german_voice directory. [[BR]] 
     171- Apply the MARY patch to the german_voice directory. [[BR]] 
    171172  patch -p1 -d . < HTS-2.0.1-demo_CMU-ARCTIC-SLT_for_Mary-3.5.0.patch 
    172173 
    … …  
    177178- Now run the VoiceImport program and follow the instructions as normal. Provide general settings for gender, locale must be "de", path to mary_base and name of the voice. 
    178179 
     180[[BR]] 
     181[[BR]] 
    179182 
    180183Marcela Charfuelan