Video Translation Software version 2.04 and later support using the Fish TTS API for voiceovers.

Deploying Fish TTS and Starting the API Service

Note: From pyvideotrans version v3.63 onwards, ensure that the fish-speech version is no less than 1.5.0, otherwise it cannot be used.

First, you need to deploy fish-speech(>=1.5.0) and start the API service. Pay attention to starting the API service, just starting the web UI, which is just opening the browser webpage interface, is not enough. The default API address is http://127.0.0.1:8080/v1/tts

For specific deployment, API startup, and other operation methods, please refer to the relevant documentation.

fish-speech tts project open source address https://github.com/fishaudio/fish-speech

Documentation address https://speech.fish.audio

Filling in the API in the Video Translation Software

After deploying and starting, confirm that the API has started, then open the video translation software, go to the upper left corner settings menu - Fish TTS, and fill in the corresponding fish TTS API in the API address. If you have not modified it, the default is http://127.0.0.1:8080/v1/tts

Next, fill in the Reference Audio and Reference Text

The reference audio is the wav audio file of the voice you want to use. The optimal audio duration is 10s-20s, clear and without background noise. Please place this audio file in the root directory of this translation software, together with sp.exe.

If you want to use the audio named 1.wav, place it in the folder where sp.exe or sp.py is located. Assuming the text content corresponding to 1.wav is "Hello, my dear friends"

Then you need to fill in

1.wav#Hello, my dear friends

If you also want to use 2.wav and want to put it in a dedicated folder, you can create a new directory in the folder where sp.exe or sp.py is located. Assuming the new directory name is wavs, and its corresponding text is "The weather is nice today, let's go swimming", then fill in

1.wav#Hello, my dear friends
wavs/2.wav#The weather is nice today, let's go swimming

If the reference text is very long, just keep the natural line breaks, do not manually add carriage returns or line breaks.

After saving, in the main interface, select FishTTS in "Dubbing Channel", and you can see the added reference audio in "Dubbing Role"

Precautions

0. Please use simple English or numbers to name the parameter audio, do not contain spaces. If you want to put all the reference audio in a folder (this folder needs to be created in the directory where sp.exe is located), use simple English or numbers for the folder name, and do not contain spaces.
1. Fish-speech supports a limited number of languages, please make sure the text is within the supported range
2. If you have added reference audio in the settings, but the corresponding role is not displayed in the main interface, please switch the dubbing channel to edgeTTS first, and then switch back to FishTTS