AllTalks TTS - Getting Output Error

[FrogChopi]

Hello, I try to use All Talks because results are pretty good ( i use the docker image ), When Substitle edit request the api it work well ( the output is generated and work well ) but it try to get it on... a linux path ( i'm on windows ) ?

'/app/outputs/output.wav'.
   à System.IO.__Error.WinIOError(Int32 errorCode, String maybeFullPath)
   à System.IO.File.InternalCopy(String sourceFileName, String destFileName, Boolean overwrite, Boolean checkHost)
   à Nikse.SubtitleEdit.Forms.Tts.TextToSpeech.GenerateParagraphAudioAllTalk(Subtitle subtitle, Boolean showProgressBar, String overrideFileName)
   à Nikse.SubtitleEdit.Forms.Tts.TextToSpeech.buttonTestVoice_Click(Object sender, EventArgs e)

It would be possible to specify the output path where subtitleedit fetches the file ( I may have missed the config )

[niksedk]

SE probably only works with this method:

1) use git to clone the repro: git clone https://github.com/erew123/alltalk_tts.git 2) use the setup bat file: atsetup.bat (choose "standalone") 3) start with "start_alltalk.bat"

Note that you can get more voices here: https://drive.google.com/file/d/1bYdZdr3L69kmzUN3vSiqZmLRD7-A3M47/view?usp=drive_link https://drive.google.com/file/d/1CPnx1rpkuKvVj5fGr9OiUJHZ_e8DfTzP/view

[FrogChopi]

I like docker because I'm a bit lazy ^^ I'll do that :) Thank you for your reply and for the great work you do.

[niksedk]

You're welcome :)

What docker command did you use? (I'm lazy too)

[FrogChopi]

• git clone https://github.com/erew123/alltalk_tts.git && cd alltalk_tts
• Personally i use an prebuilt image :

  version: '3.8'
  services:
    alltalk-tts:
      image: joshxt/alltalk_tts:cuda # you can replace by build : . if you want to built yourself
      restart: unless-stopped
      ports:
        - "7851:7851"
      volumes:
        - ./outputs:/app/outputs/
        - ./models:/app/models/
        - ./voices:/app/voices/
        - ./finetune/put-voice-samples-in-here:/app/finetune/put-voice-samples-in-here
        - ./dockerconfig.json:/app/confignew.json
      deploy:
        resources:
          reservations:
            devices:
              - driver: nvidia
                count: all
                capabilities: [ gpu ]

• docker-compose --file docker-compose-cuda.yml up -d
• Now you can just : curl -X POST "http://localhost:7851/api/tts-generate-streaming" -d "text=\"I'll be back\" is a catchphrase associated with Arnold Schwarzenegger. It was made famous in the 1984 science fiction film The Terminator." -d "voice=arnold.wav" -d "language=en" -d "output_file=myoutputfile.wav" -d "streaming=false"
• Result in ./outputs