whatmodel.sh / tts / Raon-OpenTTS-1B

Raon-OpenTTS-1B

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

Raon-OpenTTS-1B is an open-weights text-to-speech model from KRAFTON, released May 2026. Its published size is 1B parameters. Licence: cc-by-nc-4.0.

Catalogue identity notes

KRAFTON/Raon-OpenTTS-1B, trained only on open speech data (arXiv:2605.20830).

67.9±1.9comparison only: no current run option verifiedfrom 9 benchmark results across 1 source

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 Raon-OpenTTS-1B, out of 148 with a score, then more from KRAFTON.

#modelscorewhereRAMreleased
8Gemini 3.1 Flash TTS Preview Google69±2api $50.00/M chars2026-04
9dots.tts MF 2B (MeanFlow, NFE 4) dots.tts68±3local~5.2GB2026-06
10Raon-OpenTTS-1B KRAFTON68±2comparison only2026-05
11Grok TTS xAI68±3api $15.00/M chars2026-04
12FireRedTTS3 (base) FireRedTeam (Xiaohongshu)67±3local~12GB2026-08
more from KRAFTON
24Raon-OpenTTS-0.3B KRAFTON65±2comparison only2026-04

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

$ what it's good at

voice cloning▁·▁▁▁▁▁▁██▃▃ #12 of 43 * · 19% of its score · 2 tests

Beats Gemini 3.1 Flash TTS Preview here, though they rank higher overall. Trails VoxCPM2, Higgs Audio V3 TTS, dots.tts SOAR 2B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Seed-TTS test-en · speaker similarity (WavLM-large SV) lab74.9% ✎▄····▄·█▄█▄▄ beats 8/8medium · avg 53%▇▇▇▇▇ 13%typical
Seed-TTS test-en · Whisper large-v3 WER lab1.8% ✎▄···█▄▄·█▄·▄ beats 7/8avg 3.6%▇▇▇ 6.6%blunt
other tests7 not in a task · 79% of its score
testresultvs the others in its rundifficultycountswhy
CV3-Eval EN (zero-shot English) · WER lab3.9% ✎▄··▄··▄▄·█·▄ beats 6/6avg 16%▇▇▇▇▇▇▇▇ 20%typical
CV3-Eval hard-en · speaker similarity (ERes2Net SV) lab77.5% ✎▄··█··▄·█··▄ beats 6/6medium · avg 49%▇▇▇▇▇ 12%at its level
CV3-Eval hard-en · DNSMOS (predicted quality) lab3.85 / 5 ✎▄·····█·▄▄·█ beats 3/6medium · avg 2.98 / 5▇▇▇▇ 11%at its level
Raon-OpenTTS-Eval · overall speaker similarity (WavLM) lab69.5% ✎▄·█·█▄·····▄ beats 6/6medium · avg 44%▇▇▇▇ 11%at its level
CV3-Eval hard-en · WER lab6.2% ✎▄·······█·██ beats 6/6avg 26%▇▇▇▇ 10%sharp
Raon-OpenTTS-Eval · human speaker-similarity MOS (overall) lab3.70 / 5 ✎█··██······▄ beats 6/6medium · avg 2.84 / 5▇▇▇▇ 10%at its level
Raon-OpenTTS-Eval · overall WER (4 acoustic conditions) lab2.8% ✎▄··▄█····▄·█ beats 6/6avg 20%▇▇ 5.7%noisy

$ where to run it

No verified API or download recorded yet · add one

Official weights and documentation: huggingface.co/KRAFTON ↗

$ how the score was worked out · Raon-OpenTTS: Open Models and… 99%

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
Raon-OpenTTS: Open Models and Data for Robust Text-to-Speech (arXiv:2605.20830v2) · lab-reported9▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 99%CC BY 4.0
prior · pull toward the middle▇ 1%
weighted average= 67.9

$ 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