whatmodel.sh / tts / Tortoise TTS v2

Tortoise TTS v2

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Tortoise TTS v2 is an open-weights text-to-speech model from Tortoise, released April 2022. Licence: apache-2.0.

Catalogue identity notes

TorToiSe; https://huggingface.co/jbetker/tortoise-tts-v2. Submission inference details: used kv_cache. No weight commit supplied.

Model information sources

Release date ↗ · Weights licence ↗

29.2±2.3#106 of 118 text-to-speech modelsfrom 14 benchmark results across 8 sources

Score within the Vapi Humanness · Bradley–Terry rating, 2026-09-25 snapshot comparison set only.

$ how it compares

The text-to-speech models ranked either side of Tortoise TTS v2, out of 148 with a score.

#modelscorewhereRAMreleased
130StyleTTS 2 (LibriTTS) Yinghao Aaron Li et al. (Columbia University)30±2local~0.9GB2023-11
131XTTS v2 Coqui30±2local~1.9GB2023-11
132Tortoise TTS v2 Tortoise29±2local~4.5GB2022-04
133Parler-TTS Large v1 Hugging Face (Parler-TTS)29±3local~9.3GB2024-08
134WhisperSpeech v1.95 Medium (7-language) Collabora28±3local~3.2GB2024-05

see all text-to-speech models · find the ones that fit your machine

$ what it's good at

voice cloning▁·▃▁▁▁▁▁▇█▃▃ #36 of 37 * · 16% of its score · 1 test
testresultvs the others in its rundifficultycountswhy
ZS-TTS-Evaluation · English zero-shot cloning (DAPS speakers, FLORES+) · Whisper-large-v3 CER1.1% ✎█······█··██ beats 1/3avg 0.0%▇▇▇▇▇▇ 16%typical
other tests13 not in a task · 82% of its score
testresultvs the others in its rundifficultycountswhy
TTSDS2 listening test · speaker-similarity MOS (4 English domains)2.73 / 5 ✎▅·█▅▅·█▃▃▅·▅ beats 6/17easy · avg 4.03 / 5▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 41%at its level
ZS-TTS-Evaluation · English zero-shot cloning (DAPS, FLORES+) · ECAPA2 speaker similarity54.9% ✎█····▄·····▄ beats 2/3easy · avg 80%▇▇▇ 7.3%at its level
SAP voice reconstruction · human BWS reconstruction preference856.2 ✎█▄█····▄··▄▄ beats 3/7avg 995▇▇ 6.1%typical
EmergentTTS · 6-category preference16.4% ✎▅▅▃▃█▅▅▃·▃·▅ beats 4/16hard · avg 30%▇▇ 6.0%typical
EmergentTTS · v1 paper automatic MOS (WV-MOS)3.03 / 5 ✎▄·█·█▄····█▄ beats 0/8easy · avg 3.93 / 5▇▇ 5.7%blunt · at its level
EmergentTTS · 6-category WER28.6% ✎▂▄··▂▄▄··█▄▆ beats 1/16avg 16%▇▇ 3.9%blunt
TTSDS2 listening test · naturalness MOS (4 English domains)3.22 / 5 ✎▅··▃▅··▅█▅██ beats 12/17medium · avg 3.30 / 5▇ 3.5%blunt · at its level
TTSDS2 multilingual benchmark · English overall (YouTube, 2025 round)87.8% ✎▂▂█·▂·█▆▃·▄▂ beats 14/17very easy · avg 89%▇ 2.8%blunt
5 more tests
testresultvs the others in its rundifficultycountswhy
CodecMOS-Accent VCTK voice clone · human speaker-similarity MOS (S-SPK-SIM)3.70 / 5 ✎▄▄█·▄▄▄▄▄·▄█ beats 3/11easy · avg 4.20 / 5▇ 1.8%blunt · at its level
CodecMOS-Accent VCTK voice clone · human naturalness MOS (S-NAT)3.94 / 5 ✎█··▄▄██▄·▄▄▄ beats 6/11easy · avg 3.99 / 5▇ 1.6%blunt · at its level
SAP voice reconstruction · TTSDS dual-reference distributional score84.1% ✎▄····▄▄▄▄▄·█ beats 3/7very easy · avg 86%▇ 1.2%blunt
TTSDS v1 · default four-factor score87.9% ✎▂▂····▃·▃▅▂█ beats 12/14very easy · avg 86%▇ 0.5%blunt
SAP voice reconstruction · WER of reconstructed speech (Whisper turbo)15.1% ✎▄▄▄··█▄··▄·▄ beats 5/7avg 17%▇ 0.1%blunt · noisy

$ where to run it

Other ways to run it

version

  1. Run it as published · needs ~4.53 GB · from github.com/idiap
    uv run --python 3.11 --with coqui-tts --with torch --with torchaudio --with torchcodec tts --model_name tts_models/en/multi-dataset/tortoise-v2 --text "This is an example." --out_path "output.wav" --device cpu

Weights and documentation: huggingface.co/jbetker ↗

$ how the score was worked out · TTSDS2 listening test results… 45%, ZS-TTS-Evaluation (Edresson C… 23%, EmergentTTS-Eval six-category… 10% and 5 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
TTSDS2 listening test results (ttsds/listening_test)2▇▇▇▇▇▇▇▇▇▇▇▇▇ 45%MIT
ZS-TTS-Evaluation (Edresson Casanova) · zero-shot TTS metrics of XTTS, YourTTS, TorToiSe and StyleTTS 22▇▇▇▇▇▇▇ 23%MIT
EmergentTTS-Eval six-category leaderboard2▇▇▇ 10%Apache-2.0
An Evaluation Framework for TTS Voice Reconstruction (SAP dysarthric speakers; arXiv:2606.21343v1)3▇▇ 7%CC BY 4.0
EmergentTTS-Eval v1 paper (arXiv:2505.23009v1) · Table 1 automatic MOS1▇▇ 6%CC BY 4.0
CodecMOS-Accent: listening-test MOS of zero-shot TTS across English accents (arXiv:2603.14328v1)2▇ 3%CC BY 4.0
TTSDS2 multilingual benchmark results (ttsds/leaderboard_data)1▇ 3%MIT
TTSDS v1 published results1▇ 1%MIT
prior · pull toward the middle▇ 2%
weighted average= 29.2

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json