whatmodel.sh / tts / WhisperSpeech v1.95 Medium (7-language)

WhisperSpeech v1.95 Medium (7-language)

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

WhisperSpeech v1.95 Medium (7-language) is an open-weights text-to-speech model from Collabora, released May 2024. Licence: mit.

Catalogue identity notes

WhisperSpeech t2s-v1.95-medium-7lang.model + s2a-v1.95-medium-7lang.model (collabora/whisperspeech, now WhisperSpeech/WhisperSpeech; files committed 2024-05-14). Other WhisperSpeech sizes/versions are separate checkpoints.

28.0±3.3#108 of 118 text-to-speech modelsfrom 3 benchmark results across 2 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 WhisperSpeech v1.95 Medium (7-language), out of 148 with a score.

#modelscorewhereRAMreleased
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
135Glow-TTS + HiFi-GAN (VCTK, VITS study) VITS study (public implementations)28±6comparison only2021-06
136HierSpeech++ (LibriTTS-960) Korea University (Sang-Hoon Lee et al.)28±5local~2.1GB2023-11

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

$ what it's good at

other tests3 not in a task · 96% of its score
testresultvs the others in its rundifficultycountswhy
TTSDS2 listening test · speaker-similarity MOS (4 English domains)2.68 / 5 ✎▅·█▅▅·█▃▃▅·▅ beats 5/17easy · avg 4.03 / 5▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 83%at its level
TTSDS2 listening test · naturalness MOS (4 English domains)3.12 / 5 ✎▅··▃▅··▅█▅██ beats 11/17medium · avg 3.30 / 5▇▇▇ 7.0%blunt · at its level
TTSDS2 multilingual benchmark · English overall (YouTube, 2025 round)86.2% ✎▂▂█·▂·█▆▂·▄▂ beats 7/17very easy · avg 89%▇▇ 5.8%blunt

$ where to run it

Other ways to run it

  1. Run it as published · needs ~3.18 GB, NVIDIA only · from github.com/WhisperSpeech
    uv run --python 3.12 --with whisperspeech==0.8.9 --with webdataset --with torchcodec python -c "from whisperspeech.pipeline import Pipeline; pipe = Pipeline(t2s_ref='WhisperSpeech/WhisperSpeech:t2s-v1.95-medium-7lang.model', s2a_ref='WhisperSpeech/WhisperSpeech:s2a-v1.95-medium-7lang.model'); pipe.generate_to_file('output.wav', 'This is a test')"

Weights and documentation: huggingface.co/WhisperSpeech ↗

$ how the score was worked out · TTSDS2 listening test results… 90%, TTSDS2 multilingual benchmark… 6%

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▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 90%MIT
TTSDS2 multilingual benchmark results (ttsds/leaderboard_data)1▇▇ 6%MIT
prior · pull toward the middle▇ 4%
weighted average= 28.0

$ 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