Coqui TTS
Deep-learning toolkit for text-to-speech, battle-tested in research and production
Retired
Coqui TTS appears to be discontinued or shut down — company shut down Jan 2024; community-carried This page is kept as an archive record and is excluded from search engines.
About
Coqui TTS is an open-source deep-learning toolkit for text-to-speech, originally developed by the former Mozilla ML team that spun out as Coqui in 2021. It ships 1100+ pretrained models (Tacotron2, GlowTTS, YourTTS, VITS and others) plus tools to train and fine-tune your own models in any language, and dataset curation utilities. Its flagship model, XTTS v2, does zero-shot voice cloning from a ~6-second reference clip across 17 languages with cross-language cloning, and streams with sub-200ms latency. The code is MPL-2.0; the company shut down in January 2024 and the repository is now community-carried (a community fork exists at idiap/coqui-ai-TTS). Note: XTTS v2 weights are under the Coqui Public Model License — non-commercial use only.
What it can do
- 1100+ pretrained TTS models (Tacotron2, GlowTTS, SpeedySpeech, YourTTS, VITS, XTTS) in many languages
- XTTS v2: zero-shot voice cloning from a ~6s clip, 17 languages, cross-language cloning, <200ms streaming
- Train and fine-tune TTS models in any language with included recipes
- Dataset analysis and curation utilities
- Wraps external engines: Bark, Tortoise, ~1100 Fairseq MMS models
- Python API, CLI (`tts`), and Docker CPU/GPU images
Setup
- Install the Python package (verified from README):
```bash
pip install TTS
```
- List available models:
tts --list_models - Synthesize via the Python API:
```python
from TTS.api import TTS
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to("cuda")
tts.tts_to_file("Hello world", speaker_wav="my_voice.wav", language="en", file_path="out.wav")
```
- For training/fine-tuning, clone and install in dev mode:
git clone https://github.com/coqui-ai/TTS && pip install -e .[all,dev,notebooks]
NOTE: the docs quickstart page (tts.readthedocs.io) returned 404 at check time — if the exact XTTS snippet above fails, check the GitHub repo's docs folder and issues first; the project is community-maintained since Jan 2024.
Prerequisites
Python 3.9+ (PyTorch required — CUDA GPU strongly recommended for XTTS; CPU works but is much slower). Several GB of disk for model checkpoints. For voice cloning: a clean ~6-second reference recording of the target voice. Non-commercial license terms apply to XTTS v2 weights (CPML).
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: 46,110 GitHub stars (log-scaled vs catalog peers); blended with ~50,560 monthly package downloads; last push 787d ago; 0 releases/12mo; 3 open issues. Manual review: company SHUT DOWN Jan 2024 (confirmed via HN + GitHub discussion); repo stale since Aug 2024 (community fork idiap/coqui-ai-TTS); XTTS v2 weights CPML = non-commercial only; scarf.sh tracking pixel in README; docs quickstart 404'd. License: MPL-2.0.
★ 46,110 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.2/10
- Maintenance · recency, releases, issue load
- 3.9/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · 46,110 GitHub stars (log-scaled vs catalog peers)
- · blended with ~50,560 monthly package downloads
- · last push 787d ago; 0 releases/12mo; 3 open issues
- · manual review: company SHUT DOWN Jan 2024 (confirmed via HN + GitHub discussion); repo stale since Aug 2024 (community fork idiap/coqui-ai-TTS); XTTS v2 weights CPML = non-commercial only; scarf.sh tracking pixel in README; docs quickstart 404'd
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.