whatmodel.sh / tts / MMS TTS Amharic

MMS TTS Amharic

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

MMS TTS Amharic is an open-weights text-to-speech model from Meta, released May 2023. Its published size is 0.036B parameters. Licence: cc-by-nc-4.0.

Catalogue identity notes

Exact model facebook/mms-tts-amh; voice single-speaker. Dataset label Meta MMS. No API immutable build/weight SHA supplied.

Model information sources

Release date ↗ · Weights licence ↗ · Parameter count ↗

Whole VITS model: HF Hub safetensors total for the official repo (pinned revision), 36,282,672 = model.safetensors (end-to-end VITS, vocoder included). Observed 2026-10-07.

49.7±15.8#67 of 118 text-to-speech modelsfrom 1 benchmark result across 1 source* thinly tested, treat the rank as provisional

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 MMS TTS Amharic, out of 148 with a score, then more from Meta.

#modelscorewhereRAMreleased
83Addis Voice 2 Addis AI50±16comparison only
84LEMAS-TTS (w/o prosody, multilingual_grl) LEMAS Project50±6local~1.4GB2026-01
85MMS TTS Amharic Meta50±16local~0.2GB2023-05
86Chatterbox Resemble AI50±5api $25.00/M chars +local~2.9GB2025-05
87Parler-TTS Mini v0.1 Hugging Face (Parler-TTS)49±14local~2.6GB2024-04
more from Meta
144MMS TTS Turkish Meta18±4local~0.2GB2023-09

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

$ what it's good at

other tests1 not in a task · 0.2% of its score
testresultvs the others in its rundifficultycountswhy
Amharic TTS · 87-prompt CER (Chirp 3 judge)7.9% ✎▄·····▄·▄·██ beats 2/6avg 8.7%▇ 0.2%blunt · noisy

$ where to run it

Other ways to run it

  1. Run it as published · needs ~0.15 GB · from huggingface.co/facebook
    uv run --python 3.11 --with transformers --with torch --with scipy --with uroman python -c "from transformers import VitsModel, AutoTokenizer; import torch, scipy; model = VitsModel.from_pretrained('facebook/mms-tts-amh'); tokenizer = AutoTokenizer.from_pretrained('facebook/mms-tts-amh'); text = 'ሰላም፣ እንዴት ነህ?'; inputs = tokenizer(text, return_tensors='pt'); torch.set_grad_enabled(False); output = model(**inputs).waveform; scipy.io.wavfile.write('techno.wav', rate=model.config.sampling_rate, data=output.squeeze().numpy())"

Weights and documentation: huggingface.co/facebook ↗

$ how the score was worked out · Addis AI Amharic TTS benchmark 0%

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
Addis AI Amharic TTS benchmark1▇ 0%CC BY-SA 4.0
prior · pull toward the middle▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 100%
weighted average= 49.7

$ 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