IndexTTS Workflow Studio: An Enhanced Zero-Shot TTS System with Advanced Workflow Tools

👉🏻 IndexTTS Workflow Studio 👈🏻

[HuggingFace Demo] [ModelScope Demo]
[Original IndexTTS Paper] [Original IndexTTS Demos]

IndexTTS Workflow Studio builds upon the original IndexTTS (a GPT-style text-to-speech model based on XTTS and Tortoise) and adds a comprehensive user interface and workflow tools for generating, reviewing, and post-processing speech. It retains the core capabilities of the original model while providing significant enhancements for practical use cases.

This version includes major additions such as a multi-tab Gradio UI, advanced seed management, interactive audio review with speaker similarity feedback, audio concatenation, and extensive post-processing effects. See the NOTICE file for a detailed breakdown of modifications. Coming Soon: Keep an eye out for DialogueLab, a related project designed to work seamlessly with the API provided by IndexTTS Workflow Studio for creating interactive dialogues (Release expected soon!)

Author & Contact

• Author: James A Whittaker-Bent
• Email: Whittakerbent@googlemail.com
• Development Note: This Workflow Studio, featuring a comprehensive Gradio UI, audio post-processing pipeline, and FastAPI integration, represents a significant expansion developed in an intensive 3-day sprint, effectively leveraging modern AI development tools.
• Feel free to reach out regarding potential collaborations or questions about the Workflow Studio additions.

Original IndexTTS Contact (for questions about the core model):

• QQ Group: 553460296
• Discord: https://discord.gg/uT32E7KDmy

License and Attribution

This project, IndexTTS Workflow Studio, is licensed under the Apache License, Version 2.0 (see the LICENSE file).

It is based on the original IndexTTS project by Wei Deng, Siyi Zhou, Jingchen Shu, Jinchao Wang, Lu Wang (Repository: https://github.com/index-tts/index-tts), which is also licensed under Apache 2.0.

Significant modifications and additions were made by James A Whittaker-Bent. Please see the NOTICE file for detailed attribution, a summary of changes, and information regarding the licensing of the original pre-trained models (see INDEX_MODEL_LICENSE).

📣 Updates

• 2025/03/25 🔥🔥 We release the model parameters and inference code.
• 2025/02/12 🔥 We submitted our paper on arXiv, and released our demos and test sets.

🖥️ Method

The overview of IndexTTS is shown as follows.

The main improvements and contributions are summarized as follows:

• In Chinese scenarios, we have introduced a character-pinyin hybrid modeling approach. This allows for quick correction of mispronounced characters.
• IndexTTS incorporate a conformer conditioning encoder and a BigVGAN2-based speechcode decoder. This improves training stability, voice timbre similarity, and sound quality.
• We release all test sets here, including those for polysyllabic words, subjective and objective test sets.

Model Download

HuggingFace	ModelScope
Original IndexTTS Models (HuggingFace)	Original IndexTTS Models (ModelScope)

📑 Evaluation

Word Error Rate (WER) Results for IndexTTS and Baseline Models

Model	aishell1_test	commonvoice_20_test_zh	commonvoice_20_test_en	librispeech_test_clean	avg
Human	2.0	9.5	10.0	2.4	5.1
CosyVoice 2	1.8	9.1	7.3	4.9	5.9
F5TTS	3.9	11.7	5.4	7.8	8.2
Fishspeech	2.4	11.4	8.8	8.0	8.3
FireRedTTS	2.2	11.0	16.3	5.7	7.7
XTTS	3.0	11.4	7.1	3.5	6.0
IndexTTS	1.3	7.0	5.3	2.1	3.7

Speaker Similarity (SS) Results for IndexTTS and Baseline Models

Model	aishell1_test	commonvoice_20_test_zh	commonvoice_20_test_en	librispeech_test_clean	avg
Human	0.846	0.809	0.820	0.858	0.836
CosyVoice 2	0.796	0.743	0.742	0.837	0.788
F5TTS	0.743	0.747	0.746	0.828	0.779
Fishspeech	0.488	0.552	0.622	0.701	0.612
FireRedTTS	0.579	0.593	0.587	0.698	0.631
XTTS	0.573	0.586	0.648	0.761	0.663
IndexTTS	0.744	0.742	0.758	0.823	0.776

MOS Scores for Zero-Shot Cloned Voice

Model	Prosody	Timbre	Quality	AVG
CosyVoice 2	3.67	4.05	3.73	3.81
F5TTS	3.56	3.88	3.56	3.66
Fishspeech	3.40	3.63	3.69	3.57
FireRedTTS	3.79	3.72	3.60	3.70
XTTS	3.23	2.99	3.10	3.11
IndexTTS	3.79	4.20	4.05	4.01

Usage Instructions

Important Note on Models: This repository contains the code for IndexTTS Workflow Studio. The large pre-trained model files (.pth, .vocab, .model) are not included here. You must download them separately from the original IndexTTS project resources (HuggingFace or ModelScope, see Model Download section above) and place them in the checkpoints directory. These models are subject to the INDEX_MODEL_LICENSE file included in this repository.

Environment Setup

Prerequisites:

• Git
• Conda (Miniconda or Anaconda)
• Python 3.10

1. Clone this repository:

           # Replace with your actual repository URL after uploading to GitHub
           git clone https://github.com/JaySpiffy/IndexTTS-Workflow-Studio.git
           cd IndexTTS-Workflow-Studio

2. Create Conda Environment:

   conda create -n index-tts-studio python=3.10
   conda activate index-tts-studio

3. Install Dependencies:

   pip install -r requirements.txt

   • FFmpeg: You also need FFmpeg installed.
     • Linux (Debian/Ubuntu): sudo apt-get update && sudo apt-get install ffmpeg
     • macOS (using Homebrew): brew install ffmpeg
     • Windows (using Chocolatey): choco install ffmpeg or download from the official FFmpeg website and add to your PATH.
   • Windows pynini Note: If you encounter errors installing pynini on Windows (Failed building wheel for pynini), install it via conda before running pip install -r requirements.txt:

     conda install -c conda-forge pynini==2.1.5
     # Then run: pip install -r requirements.txt

4. Download Models: Download the required model files (bigvgan_discriminator.pth, bigvgan_generator.pth, bpe.model, dvae.pth, gpt.pth, unigram_12000.vocab) from the links in the "Model Download" section above and place them into the checkpoints/ directory within this project. You can use wget or huggingface-cli:

   • Using huggingface-cli:

     # Recommended for China users: export HF_ENDPOINT="https://hf-mirror.com"
     huggingface-cli download IndexTeam/Index-TTS \
       bigvgan_discriminator.pth bigvgan_generator.pth bpe.model dvae.pth gpt.pth unigram_12000.vocab \
       --local-dir checkpoints --local-dir-use-symlinks False

   • Using wget:

     wget https://huggingface.co/IndexTeam/Index-TTS/resolve/main/bigvgan_discriminator.pth -P checkpoints
     wget https://huggingface.co/IndexTeam/Index-TTS/resolve/main/bigvgan_generator.pth -P checkpoints
     wget https://huggingface.co/IndexTeam/Index-TTS/resolve/main/bpe.model -P checkpoints
     wget https://huggingface.co/IndexTeam/Index-TTS/resolve/main/dvae.pth -P checkpoints
     wget https://huggingface.co/IndexTeam/Index-TTS/resolve/main/gpt.pth -P checkpoints
     wget https://huggingface.co/IndexTeam/Index-TTS/resolve/main/unigram_12000.vocab -P checkpoints

Running the Web Demo (Workflow Studio)

python webui.py

Open your browser and visit http://127.0.0.1:7860 (or the URL provided in the terminal) to access the Gradio interface.

For detailed instructions on using the UI features, please see the Gradio UI Guide.

Running the API Server

# Ensure you are in the project root directory with the conda environment activated
uvicorn tts_api:app --host 0.0.0.0 --port 8000

The API will be available at http://localhost:8000. See tts_api.py for endpoint details (e.g., /synthesize).

Basic Inference (Sample Code)

from indextts.infer import IndexTTS
tts = IndexTTS(model_dir="checkpoints",cfg_path="checkpoints/config.yaml")
voice="reference_voice.wav"
text="大家好，我现在正在bilibili 体验 ai 科技，说实话，来之前我绝对想不到！AI技术已经发展到这样匪夷所思的地步了！比如说，现在正在说话的其实是B站为我现场复刻的数字分身，简直就是平行宇宙的另一个我了。如果大家也想体验更多深入的AIGC功能，可以访问 bilibili studio，相信我，你们也会吃惊的。"
tts.infer(voice, text, output_path)

Dialogue Generation Prompt

For those interested in the methodology, the detailed system prompt used to guide AI (like Cline) in generating the formatted dialogue scripts for the TTS engine can be found in the file DIALOGUE_GENERATION_PROMPT.md.

Acknowledge (Core Dependencies)

1. tortoise-tts
2. XTTSv2
3. BigVGAN
4. wenet
5. icefall

📚 Citation (Original IndexTTS Paper)

🌟 If you find our work helpful, please leave us a star and cite our paper.

@article{deng2025indextts,
  title={IndexTTS: An Industrial-Level Controllable and Efficient Zero-Shot Text-To-Speech System},
  author={Wei Deng, Siyi Zhou, Jingchen Shu, Jinchao Wang, Lu Wang},
  journal={arXiv preprint arXiv:2502.05512},
  year={2025}
}