Moonshine
Very low latency speech to text, intent recognition, and text to speech, for building voice agents and interfaces
Moonshine scores 6.7/10 (Solid) in Voice & Realtime.
Language C++ License Custom (see repo) Stars 8.6k★ Type model Self-host Yes Repo usefulsensors/moonshine Homepage https://github.com/usefulsensors/moonshine
Quality
Solid 6.7
Adoption 48
Activity 63
Maturity 89
Community 43
Capability 90
Show the math
Overall 6.7/10
value = 6.7/10
score = value
Adoption 5.4/10
value = 48.4/100
score = 1 + 9 * value / 100
Activity 6.6/10
value = 62.7/100
score = 1 + 9 * value / 100
Maturity 9/10
value = 88.6/100
score = 1 + 9 * value / 100
Community 4.9/10
value = 43.2/100
score = 1 + 9 * value / 100
Capability 9.1/10
value = 89.9/100
score = 1 + 9 * value / 100
Key metrics
8.6k★ GitHub stars
104/90d Recent commits
3 devs/90d Recent contributors
10.07% avg WER WER, Open ASR Leaderboard 8-dataset average (Moonshine Base) Moonshine Base averages 10.07% WER across the 8 OpenASR Leaderboard datasets (LibriSpeech-clean 3.23%, -other 8.18%), beating Whisper base.en (10.32%). Reported as 100-WER of the average. Source: arXiv 2410.15608 (https://arxiv.org/html/2410.15608v2).
1.7y Project age
active Last commit
Gotchas
No gotchas documented yet. Contribute one if you know a constraint we missed.
Labels
- No license
No OSS license
No recognized open-source license is declared; using it commercially is legally ambiguous.
- Self-hosted
Self-hosted
You deploy and operate it yourself; there is no hosted option here.
asr on-device edge low-latency streaming