whatmodel.sh / tts / Grad-TTS + HiFi-GAN (10 steps)

Grad-TTS + HiFi-GAN (10 steps)

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

Grad-TTS + HiFi-GAN (10 steps) is an open-weights text-to-speech model from Grad-TTS authors / Huawei Noah, released May 2021. Its published size is 0.0148B parameters.

Catalogue identity notes

Original Grad-TTS acoustic checkpoint with HiFi-GAN; 10 reverse-diffusion steps, temperature 1.5.

Model information sources

· Parameter count ↗

Grad-TTS acoustic model only: OUR SUM of the makers paper Table 2 figures (encoder + duration predictor 7.2M, diffusion decoder 7.6M). The HiFi-GAN vocoder of this two-stage system is not included.

64.8±23.3#1 of 1 text-to-speech modelsfrom 1 benchmark result across 1 source* thinly tested, treat the rank as provisional

Score within the Grad-TTS 2021 LJ Speech human MOS comparison set only.

$ how it compares

Scored text-to-speech models released closest to Grad-TTS + HiFi-GAN (10 steps), out of 148 with a score.

#modelscorewhereRAMreleased
81Azure Ameha Neural Microsoft50±16comparison only2021-10
82Azure Mekdes Neural Microsoft50±16comparison only2021-10
124VITS (VCTK multi-speaker) Kakao Enterprise (Jaehyeon Kim et al.)34±5local~0.2GB2021-06
135Glow-TTS + HiFi-GAN (VCTK, VITS study) VITS study (public implementations)28±6comparison only2021-06
142Tacotron 2 + HiFi-GAN (VCTK, VITS study) VITS study (public implementations)19±6comparison only2021-06

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

$ what it's good at

other tests1 not in a task · 56% of its score
testresultvs the others in its rundifficultycountswhy
Grad-TTS 2021 LJ Speech human MOS lab4.38 / 5 ✎█·····█···██ beats 3/3easy · avg 4.17 / 5▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 56%typical

$ where to run it

Other ways to run it

  1. Run it as published · needs ~0.12 GB, NVIDIA only · from github.com/huawei-noah (official)
    git clone https://github.com/huawei-noah/Speech-Backbones && cd Speech-Backbones/Grad-TTS && pip install -r requirements.txt && cd model/monotonic_align && python setup.py build_ext --inplace && cd ../.. && sed -i 's/^pe_scale = 1000/pe_scale = 1/' params.py && python inference.py -f resources/filelists/synthesis.txt -c checkpts/grad-tts-old.pt -t 10

Official weights and documentation: github.com/huawei-noah ↗

$ how the score was worked out · Grad-TTS: original ICML 2021… 56%

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
Grad-TTS: original ICML 2021 listening study · lab-reported1▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 56%CC-BY-4.0
prior · pull toward the middle▇▇▇▇▇▇▇▇▇▇▇▇▇ 44%
weighted average= 64.8

$ 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