Changes between Version 4 and Version 5 of HMMVoiceCreationMary4.0


Ignore:
Timestamp:
09/10/09 18:13:58 (16 years ago)
Author:
marcela_charfuelan
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • HMMVoiceCreationMary4.0

    v4 v5  
    7575 
    7676''' 
    77 === II) Data preparation: === 
     77=== II) Data preparation (If you want to create a HMM voice in other language please see the section V or VI below): === 
     78 
    7879''' 
    7980 
    … …  
    111112Tutorial: http://mary.opendfki.de/wiki/VoiceImportToolsTutorial 
    112113 
    113 If you want to create another voice in German or English please see the section V below. 
    114  
    115114Please remember that whenever you are in doubt about the settings of a particular component you can check its corresponding help for a description of the meaning 
    116115(and possible values) of each variable. 
    117116 
     117After starting the voiceimport tools check the global settings of the voice, make sure that the allophones file is provided and exist: 
     118{{{ 
     119de.alophonesSet = MARY_PATH/lib/modules/xx/lexicon/allophones.xx.xml  (where xx is the corresponding language) 
     120}}} 
     121 
    118122'''5-''' Run the HMMVoiceDataPreparation of the HMM Voice Trainer group to check if text, wav and data/raw files are available and in the correct paths. 
    119123If just data/raw is provided, the program will do the conversion.  If no text files are available but data/utts in festival format, the program will do the conversion as well. 
    … …  
    127131   EHMMLabeler.ehmm  = ../festvox/src/ehmm/bin/ 
    128132}}} 
     133The result of this step is a '''lab''' directory. 
    129134 
    130135'''8-''' Run the LabelPauseDeleter component of the Automatic Labeling group. Please use the settings editor of this component to set the variable: 
    … …  
    135140'''9-''' CHECK: Here do we need to run the TranscriptionAligner???  This program will create the '''allophones''' directory! what is the difference between allophones and prompt_allophones??? 
    136141 
    137 '''10-''' Run the PhoneUnitLabelComputer component of the Labels and Pause Correction group. This procedure will create a '''phonelab''' directory.  
    138  
    139 '''11-''' FeatureSelelection creates '''mary/features.txt''', it requires the MARY server running. 
     142'''10-''' Run the PhoneUnitLabelComputer component of the Labels and Pause Correction group. This procedure has as input the '''lab''' directory created with the EHMMLabeler and will create as an output the '''phonelab''' directory.  
     143 
     144'''11-''' FeatureSelelection creates '''mary/features.txt''', it requires the MARY server running. Select here all the features and save the file. 
    140145 
    141146'''12-''' Run the PhoneUnitFeatureComputer component of the Feature Extraction group to extract context feature vectors from the text data. This procedure will create a '''phonefeatures''' directory. For running this component the MARY server should be running as well.  
    … …  
    181186   ... 
    182187}}} 
    183 If you are not sure about using other features, use just the first four, delete the others and save the file. 
    184  
    185 '''16-''' Run the HMMVoiceMakeData component of the HMM Voice trainer group to run the HTS procedure "make data". This procedure require the following files, please use the settings editor of this component to set the variables: 
     188If you are not sure about using other features, use the first four, delete the others and save the file. 
     189 
     190'''16-''' Run the HMMVoiceMakeData component of the HMM Voice trainer group to run the HTS procedure "make data". This procedure require the following files: 
    186191{{{ 
    187192   HMMVoiceMakeData.allophonesFile   = allophones.en_US.xml  # allophones set (language dependent) 
    … …  
    189194}}} 
    190195 
    191 The allophones set file is language dependent, an example for US English can be found in openmary/lib/modules/en/us/lexicon/allophones.en_US.xml[[BR]] 
    192 The hmmFeatures.txt is the file created in step 16 and contains additional context features, appart from phone and phonological features, used for training HMMs.[[BR]] 
     196The allophones set file is language dependent, it can be found in MARY_PATH/lib/modules/en/us/lexicon/allophones.en_US.xml[[BR]] 
     197The hmmFeatures.txt is the file created in step 15 and contains additional context features, apart from phone and phonological features, used for training HMMs.[[BR]] 
    193198 
    194199The HMMVoiceMakeData procedure is similar to the original HTS scripts with additional sections for calculating strengths, Fourier magnitudes (for mixed excitation), global variance and composing training data files from mgc, lf0, str and mag files. This component will execute in the data/ directory:  
    … …  
    205210  HMMVocieMakeData.makeCMPMARY   =  1 
    206211}}} 
    207 all the other variables in 0, and run again the component. (In this case you need to run as well makeCMPMARY because you need to compose again the vectors mgc+lf0+str+mag). 
    208  
    209 The 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. 
     212all the other variables in 0, and run again the component. (In this case you need to run as well make CMPMARY because you need to compose again the vectors mgc+lf0+str+mag). 
     213 
     214The procedures can be repeated manually as well, going to the data directory and running "make data" or "make str", as it is normally done with the original HTS scripts. 
    210215 
    211216NOTE: the Makefile in data/ includes a gv: section to calculate global variance files. In MARY, these files are generated little endian and contain a header of size one short to indicate the size of the vectors it contains. 
    … …  
    227232''' 
    228233 
    229 '''18-''' Run the HMMVoiceInstaller component of the Install Voice group. The default setting values of this component are already fixed for the HTS-demo_CMU-ARCTIC-SLT voice. If you are training other voice  please use the settings editor of this component to set: 
    230 {{{ 
    231    
    232   HMMVoiceInstaller.FeaFile     =  phonefeatures/xx.pfeats 
    233                                    this is an example of a CONTEXTFEATURES file in MARY format for synthesise during start-up.  
     234'''18-''' Run the HMMVoiceInstaller component of the Install Voice group. The default setting values of this component are already fixed for the HTS-demo_CMU-ARCTIC-SLT voice. Some settings of the voice can be changed here, for example: 
     235{{{ 
    234236  HMMVoiceInstaller.useMixExc   =  true 
    235237                                   set this variable to true if using mixed excitation 
    … …  
    257259 
    258260For creating a new German voice it is necessary: [[BR]] 
     261  * NLP components for German, those should be available with MARY 4.0  
    259262  * a wav or raw directory with the speech files you will use for training the German voice. [[BR]] 
    260263  * transcriptions of the files, one text file per speech file, or transcriptions in festival format if available. [[BR]] 
    … …  
    262265Then we use as a base the original HTS-demo_CMU-ARCTIC-SLT directory: 
    263266 
    264 - Download and un-zip, un-tar the HTS-demo_CMU-ARCTIC-SLT for HTS-2.1 
    265  
    266 - Rename this directory as your new voice name, for example german_voice, and delete the directories data/raw and data/utt. 
     267- Download and un-zip, un-tar the HTS-demo_CMU-ARCTIC-SLT for HTS-2.1  
     268  Since English is not going to be used then the demo without data can be downloaded (http://hts.sp.nitech.ac.jp/archives/2.1/HTS-demo_CMU-ARCTIC-SLT-wodata.tar.bz2), otherwise delete the directories data/raw and data/utts from the original demo. [[BR]] 
     269 
     270- Rename this directory as your new voice name, for example german_voice, '''make sure that the directories data/raw and data/utt are deleted.''' 
    267271 
    268272- Apply the MARY patch to the german_voice directory. [[BR]] 
    … …  
    273277- Move your transcription files to this directory, if you have a text directory containing the transcription of each file in separate files, this should be copied in the current directory (german_voice/text). If you have transcriptions in festival format please copy this directory in the data/utts directory (german_voice/data/utts/). 
    274278 
    275 - Now run the VoiceImport program and follow the instructions as normal. Provide general settings for: 
     279- Now run the VoiceImport program and follow the instructions as normal from step 4. Provide general settings for: 
    276280{{{ 
    277281   db.gender    =  male  (or female) 
    … …  
    287291If you are creating a voice in other language you will need to specify: 
    288292 
    289 - '''Minimal NLP components''': if you are creating a new voice from scratch, for example following the steps in [http://mary.opendfki.de/wiki/NewLanguageSupport], you will need to create Minimal NLP components for the new language. The MARY server should contain minimal NLP components for extracting context features in the new language ('''phonefeatures''' directory). 
     293- '''Minimal NLP components''': if you are creating a new voice from scratch, for example following the steps in [http://mary.opendfki.de/wiki/NewLanguageSupport], you will need to create Minimal NLP components for the new language. These minimal components are necessary to run the MARY server in the new language and extract context features ('''phonefeatures''' directory). 
    290294 
    291295- '''Phoneme set''':  contained in MARY_BASE/lib/modules/xx/lexicon/allophones.xx.xml , where xx corresponds to the new language. 
    292296 
    293 - '''Extra context features''': during training of HMMs phones, phonological features and extra features are required. Phones and phonological features are specified in the allophones xml file. The extra features can be specified in the '''mary/hmmFeatures.txt''' file (step 15).  
    294  
     297- After creating the minimal components, follow the steps in section V.  
    295298 
    296299[[BR]] 
    … …  
    298301 
    299302Marcela Charfuelan[[BR]] 
    300 Fri Sep  4 18:03:25 CEST 2009 
     303Thu Sep 10 18:14:12 CEST 2009