pyVideoTrans

Appearance

TTS Channels: Index-TTS / Spark-TTS / VoxCPM / FireRed3-TTS ​

Starting from v3.68, these TTS services share the same settings interface. Simply enter the WebUI URL after starting the corresponding TTS service (usually http://127.0.0.1:7860 by default), then select the service from the "TTS Channel" dropdown menu on the main interface.

This document covers the deployment and integration methods for four open-source TTS services. They all provide APIs via Gradio WebUI, and their configuration processes are essentially the same.

1. Overview of Channels ​

ChannelDeveloperSupported LanguagesDefault AddressOfficial Project
Index-TTSBilibili Open SourceChinese, Englishhttp://127.0.0.1:7860https://github.com/index-tts/index-tts
FireRed3-TTSXiaohongshu24 Languageshttp://127.0.0.1:7860https://github.com/jianchang512/FireRedTTS3 (Modified version with webui.py added for integration)
VoxCPM-TTSModelBest30+ Languageshttp://127.0.0.1:7860https://github.com/OpenBMB/VoxCPM
Spark-TTSSparkAudioChinese, Englishhttp://127.0.0.1:7860https://github.com/SparkAudio/Spark-TTS

The integration workflow is identical for all channels: Launch WebUI → Enter the address in pyVideoTrans → Select the channel → Configure reference audio → Start using.

2. Prerequisites ​

Before using these TTS services, please make sure you meet the following requirements:

  1. pyVideoTrans Version: v3.68 or higher
  2. Hardware Requirements: An NVIDIA GPU (for GPU acceleration) is recommended. Some services also support running on CPU (though much slower).
  3. Network Environment: The first run may require downloading models from HuggingFace. Users in certain regions may need a proxy or configured mirror.
  4. Python Environment: If deploying from source code, Python 3.10+ and Git are required.

Index-TTS ​

Index-TTS is an open-source TTS service by Bilibili. It supports Chinese and English voice cloning with excellent sound quality.

Windows All-in-One Package ​

Source Code Deployment ​

Please refer to the official documentation: https://github.com/index-tts/index-tts

Starting WebUI ​

After launching, the default address is http://127.0.0.1:7860. The configuration process is identical to F5-TTS.

Note: Only the official Index-TTS WebUI is supported. If you use a third-party package, please overwrite the third-party webui.py with the webui.py from the official repository; otherwise, it may fail to work.

FireRed3-TTS (Xiaohongshu TTS3) ​

Supported Languages: Arabic · Cantonese · Chinese · Czech · Dutch · English · French · German · Greek · Hindi · Indonesian · Italian · Japanese · Korean · Polish · Portuguese · Romanian · Russian · Spanish · Thai · Turkish · Ukrainian · Vietnamese

The official repository does not include a WebUI and cannot be directly integrated. Therefore, a modified version with an added webui.py is provided at: https://github.com/jianchang512/FireRedTTS3

Installation & Deployment: ​

  • Requires an NVIDIA GPU with a configured CUDA 12.9 environment.
git clone https://github.com/jianchang512/FireRedTTS3
cd FireRedTTS3
pip install -r requirements.txt
  • Start webui.py: python webui.py. If using Google Colab, add the --share argument: python webui.py --share
  • Start webuiall.py: python webuiall.py. If using Google Colab, add the --share argument: python webuiall.py --share

After launching, copy the address (default local: http://127.0.0.1:7860) and paste it into pyVideoTrans Menu -> TTS Settings -> Index/VoxCPM/Spark/FireRed3 under FireRed3-TTS URL.

VoxCPM-TTS ​

VoxCPM-TTS is an open-source TTS service by ModelBest. It supports over 30 languages, including multiple Chinese dialects.

Supported Languages ​

  • International Languages (30+): Arabic, Burmese, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Indonesian, Italian, Japanese, Khmer, Korean, Lao, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Tagalog, Thai, Turkish, Vietnamese
  • Chinese Dialects: Mandarin, Sichuanese, Cantonese, Wu (Shanghainese), Northeastern, Henan, Shaanxi, Shandong, Tianjin, Minnan

Deployment ​

Deploy from source code following the official project documentation: https://github.com/OpenBMB/VoxCPM

After starting the WebUI, the default address is http://127.0.0.1:7860. The configuration process is identical to F5-TTS.

Spark-TTS ​

Spark-TTS is an open-source TTS service by SparkAudio, supporting Chinese and English voice cloning.

Deployment ​

Please deploy from source code following the official project documentation: https://github.com/SparkAudio/Spark-TTS

After starting the WebUI, the default address is http://127.0.0.1:7860. The configuration process is identical to F5-TTS.

Reference Audio Configuration ​

Reference audio is configured under Menu → TTS Settings → Set Reference Audio.

Steps ​

  1. Open the "Set Reference Audio" settings window.
  2. In the "Reference Audio" text box, enter the details in the following format:
AudioFileName#MatchingTextInTheAudio
  1. Place the reference audio file into the f5-tts folder inside the pyVideoTrans installation directory (if the folder does not exist, please create it manually).

Example ​

Suppose you have an audio file named nverguo.wav, and the spoken words in the audio are "女儿国王说话". You would enter:

nverguo.wav#女儿国王说话

Place reference audio into the f5-tts folder inside pyVideoTrans

Reference audio filename and matching text

Reference Audio Requirements ​

ItemRequirement
FormatWAV format (recommended); MP3 and other formats are also supported
Duration3–12 seconds (F5-TTS maximum is 12 seconds; longer audio will be automatically trimmed)
ContentClear pronunciation without background noise
TextMust match the spoken words in the audio file exactly

Common Errors and Notes ​

1. Do Not Close the Terminal Window ​

While using the API, you may close the WebUI page in your browser, but do NOT close the terminal/command prompt window running the TTS service.

Do not close this window, otherwise the API cannot be accessed

2. Cannot Switch Models Dynamically ​

You cannot switch models on the fly. You must manually modify the code/settings and restart the WebUI.

3. HuggingFace Connection Timeout ​

If you encounter an error similar to this:

requests.exceptions.ConnectTimeout: HTTPSConnectionPool(host='huggingface.co', port=443): Max retries exceeded

This is a network issue. Please check your network environment to ensure you can access huggingface.co properly.

4. Reference Audio Duration Limit ​

A duration of 3 to 10 seconds is recommended.