Voice building

[elnursamedzade]

Hello. Please, can you send me a build.gradle sample? I want to understand what I'm doing wrong. I am a visually impaired user. I am using Nvda screen reader. I want to use my own voice as a synthesizer. Please feel free to contact me privately. I need your help a lot. My e-mail address is elnur.samedzade@gmail.com Thank you.

[psibre]

Thanks for the context! We're in the process of updating our voices to work with the latest code, but it takes some time. I'll try to prioritize marytts/voice-dfki-ot#3, which will hopefully help you get more experience with how it's all supposed to work. =)

[elnursamedzade]

Thank you very much! Isn't there any way I can add my own data to the voice-dfki-ot voice right now? What codes do I need to change in the build.gradle file? I am replacing wav files with my own voice but md5 gives encryption error. I don't want to leave Marytts. I've been fighting for days. I prepared my own text files, recorded my voice. I don't want to leave now.

[psibre]

There's some fairly complicated build logic in the current build script for voice-dfki-ot: https://github.com/marytts/voice-dfki-ot/blob/master/build.gradle#L32-L116

This was intended to be streamlined by using a common plugin, like in marytts/voice-enst-camille#5, but back at the end of 2018, there wasn't enough time, and then I switched jobs. I only recently started to free up enough time to help maintain MaryTTS a bit more here and there.

TL;DR

1. You can wait another day or two, then the voice-dfki-ot project should be updated and easier to understand, or

2. You can hack the voice-dfki-ot build logic on your own:

   1. Copy your voice data (wav, text, and lab files) into custom directories called my_wav_files, my_text_files, and my_lab_files, respectively
   2. Use Gradle v4.10.3
   3. Patch the build logic by adapting something like this:

      
      diff --git a/build.gradle b/build.gradle
      index a6d21fc..13ebc60 100644
      --- a/build.gradle
      +++ b/build.gradle
      @@ -8,7 +8,7 @@ buildscript {
      }

   plugins {

   • id 'de.dfki.mary.voicebuilding-legacy' version '5.2.2'
   • id 'de.dfki.mary.voicebuilding-legacy' version '5.3.2' id 'de.undercouch.download' version '3.1.1' }

@@ -115,8 +115,17 @@ task extractWav { } }

+wav.srcFiles = fileTree('my_wav_files').include('*.wav') + +generateAllophones.srcDir = file('my_text_files') + +task copyLabFiles(type: Copy) {

• from 'my_lab_files'
• into "$buildDir/lab" +}

• legacyInit {

• dependsOn extractLab, extractText, extractWav
• dependsOn wav, copyLabFiles def basenameListFile = file("$buildDir/basenames.lst") outputs.files basenameListFile doLast { @@ -124,37 +133,8 @@ legacyInit { fileTree(wav.destDir).collect { it.name - '.wav' }.minus([
• 'ot0044',
• 'ot0100',
• 'ot0103',
• 'ot0256',
• 'ot0282',
• 'ot0291',
• 'ot0307',
• 'ot0323',
• 'ot0370',
• 'ot0453',
• 'ot0497',
• 'ot0538',
• 'ot0569',
• 'ot0600',
• 'ot0662',
• 'ot0673',
• 'ot0745',
• 'ot0746',
• 'ot0774',
• 'ot0791',
• 'ot0818',
• 'ot0853',
• 'ot0882',
• 'ot0883',
• 'ot0893',
• 'ot0918',
• 'ot0944',
• 'ot1073',
• 'ot1106',
• 'ot1132',
• 'ot1159'
• // list here basenames with invalid data
• // in case of processing errors ]).each { basenameList.writeLine it }

  
  But if you run into problems here, please be patient until we update the build logic properly.

[elnursamedzade]

Hello mr Psibre. Unfortunately, it did not happen. I will be patient and wait for your return. I hope one day I will be a marytts user and leave happy here. I wish you good work. Please keep me updated. Best regards.

[psibre]

@elnursamedzade Can we close this issue and continue discussion in marytts/voice-dfki-ot#3?