Open-source voice AI family from Microsoft, with long-form speech recognition (ASR), multi-speaker text-to-speech (TTS), and real-time streaming TTS models.
54.9K6.2KPythonMITUpdated 12 days ago
index-tts/index-tts
A zero-shot text-to-speech system that clones a voice from one reference audio clip, with emotion, speed and pronunciation control.
Open-source alternative to:ElevenLabs
- Stars
- 24,604
- Forks
- 2,908
- Open issues
- 413
- Last commit
- 21 days ago
What is it?
IndexTTS is a zero-shot text-to-speech system that clones a speaker's voice from a single reference audio clip. The current release, IndexTTS-2.5, generates speech in Chinese, English, Japanese, Spanish and Arabic, and can carry over the voice of a reference speaker across languages.
Beyond voice cloning, it offers control over emotion (via a separate emotional reference clip, an 8-dimension emotion vector, or emotion inferred from text), emotion intensity, and speaking speed through a duration factor from 0.5x to 2.0x. Pronunciation can be steered with Pinyin for Chinese, CMU phonemes for English and Kana for Japanese.
It runs through a Python API, a Gradio-style web demo launched with uv, and a vLLM recipe for serving. Half-precision (FP16/BF16) and optional DeepSpeed acceleration are supported on NVIDIA GPUs, and the project lists its model weights on HuggingFace and ModelScope.
Key features
- 01Zero-shot voice cloning from a single reference audio clip.
- 02Cross-lingual synthesis across Chinese, English, Japanese, Spanish and Arabic.
- 03Emotion control via reference audio, emotion vectors, or text.
- 04Speaking speed control from 0.5x to 2.0x duration.
- 05Pronunciation control with Pinyin, CMU phonemes and Japanese Kana.
- 06Includes a WebUI, Python API and vLLM serving option.
Use cases
- Generating voiceovers or narration in a specific voice from a short sample.
- Producing multilingual speech that keeps one speaker's timbre across languages.
- Adding controllable emotion and pacing to synthetic dialogue.
- Correcting pronunciation of Chinese characters, English words or Japanese kanji in generated speech.
Quick start
- 1
Install uv package manager
pip install -U uv - 2
Clone the repository
git clone https://github.com/index-tts/index-tts.git && cd index-tts - 3
Install dependencies with all extras
uv sync --all-extras - 4
Download the IndexTTS-2.5 model from HuggingFace
hf download IndexTeam/IndexTTS-2.5 --local-dir=checkpoints
Things to know
- Requires NVIDIA CUDA Toolkit 12.8 or newer for CUDA installs.
- DeepSpeed may be difficult to install on Windows.
- Model weights must be downloaded separately before use.
- Released under a license that is not a standard identifier, so check its terms before commercial use.
Topics
FAQ
- What is index-tts?
- A zero-shot text-to-speech system that clones a voice from one reference audio clip, with emotion, speed and pronunciation control. IndexTTS is a zero-shot text-to-speech system that clones a speaker's voice from a single reference audio clip. The current release, IndexTTS-2.5, generates speech in Chinese, English, Japanese, Spanish and Arabic, and can carry over the voice of a reference speaker across languages.
- Is index-tts free to use? What license does it have?
- index-tts is open source under the Other license. Check the license terms before using it in commercial products.
- Is index-tts still maintained?
- Actively maintained. The last commit to index-tts was on Sep 29, 2026.
- What can I use index-tts for?
- Generating voiceovers or narration in a specific voice from a short sample. Producing multilingual speech that keeps one speaker's timbre across languages. Adding controllable emotion and pacing to synthetic dialogue. Correcting pronunciation of Chinese characters, English words or Japanese kanji in generated speech.
- How do I install index-tts?
- Install uv package manager: pip install -U uv Clone the repository: git clone https://github.com/index-tts/index-tts.git && cd index-tts Install dependencies with all extras: uv sync --all-extras Download the IndexTTS-2.5 model from HuggingFace: hf download IndexTeam/IndexTTS-2.5 --local-dir=checkpoints
- Where can I get index-tts?
- The source code and installation instructions are on GitHub: https://github.com/index-tts/index-tts
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Python inference library for Kokoro-82M, an open-weight text-to-speech model with 82 million parameters.
9.5K1.1KJavaScriptApache-2.0Updated last year
Soprano is an ultra-lightweight, on-device text-to-speech model with CUDA, CPU and MPS support and streaming output.
1.6K138PythonApache-2.0Updated 9 months ago
Open-source text-to-speech system with voice cloning, natural-language emotion control and multi-speaker generation, built on LLM-style models.
33.3K2.9KPythonOtherUpdated 6 days ago
Open-source text-to-speech model family by Resemble AI, with voice cloning from a short reference clip across English and many other languages.
27.1K3.6KPythonMITUpdated 3 months ago
NVIDIA's open-source toolkit for building, fine-tuning and deploying speech AI models, covering speech recognition (ASR), text-to-speech (TTS) and speech LLMs.
18.6K3.7KPythonApache-2.0Updated 10 hours ago