Fish-TTS / Fish-Speech Voice Channel

What is it?

Fish-Speech is an open-source voice synthesis project that supports voice cloning — it can generate speech in a similar voice using a reference audio sample. Ideal for users who need a specific voice for dubbing.

Prerequisites:

• You need to deploy and run the Fish-Speech API service
• pyVideoTrans v3.63+ is required
• Fish-Speech version 1.5.0 or higher

Important: Starting from v3.63, pyVideoTrans requires Fish-Speech version 1.5.0 or higher. Earlier versions will not work.

Step 1: Deploy Fish-Speech

Get the Project

• Repository: https://github.com/fishaudio/fish-speech
• Documentation: https://speech.fish.audio

Start the API Service

1. Follow the Fish-Speech documentation to complete deployment
2. You must start the API service — launching only the WebUI is not enough
3. Default API address: http://127.0.0.1:8080/v1/tts

Note: Make sure the API service is running before proceeding with configuration.

Step 2: Configure the API in pyVideoTrans

1. Open pyVideoTrans
2. Go to Menu → Settings → Fish TTS
3. Enter the Fish TTS API address in the API Address field
4. Default address: http://127.0.0.1:8080/v1/tts

Step 3: Configure Reference Audio

Fish-Speech uses reference audio to clone voices. You need to provide a reference audio file along with its corresponding text.

Format

audio_file_path#corresponding text

Examples

Example 1: Single reference audio

If you have a 1.wav file in the same folder as sp.exe or sp.py, with the corresponding text "Hello, my dear friends", enter:

1.wav#Hello, my dear friends

Example 2: Multiple reference audio files

If you want to use multiple reference audio files, create a folder (e.g., wavs) in the sp.exe directory, then enter:

1.wav#Hello, my dear friends
wavs/2.wav#Nice weather today, let's go for a swim

Reference Audio Requirements

• Format: WAV format
• Duration: 10-20 seconds is optimal
• Quality: Clear, no background noise
• Location: Place in the same folder as sp.exe or sp.py
• Naming: Use simple English or numeric names without spaces

Tip: If the reference text is very long, just let it wrap naturally — don't add manual line breaks.

Step 4: Use Fish-TTS for Dubbing

1. Save the settings, then return to the main interface
2. Select FishTTS from the Voice Channel dropdown
3. Your added reference audio will appear in the Voice dropdown
4. Select a voice and start dubbing

Important Notes

1. File naming: Use simple English or numeric names for reference audio files and folders — no spaces
2. Language support: Fish-Speech has limited language support — make sure the text is within supported languages
3. Voices not showing: If you added reference audio in settings but it doesn't appear on the main interface, switch the voice channel to edgeTTS first, then switch back to FishTTS
4. API must be running: Just opening the browser interface is not enough — the API service must be active

Common Errors

Error	Cause	Solution
"This channel needs deployed and started"	Fish-Speech is not running	Ensure the API service is running
Connection refused	Incorrect API address	Check the API address and port
Voices not appearing	Caching issue	Switch to another channel and back
Generated voice doesn't sound like the reference	Poor reference audio quality	Use clearer, longer reference audio