Moonshine open-weights STT models
Hacker ApprovedRanked 4 of 25 speech recognition models by recommenders
Open-weights speech-to-text models praised for fast streaming latency and accuracy close to larger models, running on small hardware. Critics say Parakeet beats them on accuracy and speed.
Open-weights speech-to-text models praised for fast streaming latency and accuracy close to larger models, running on small hardware. Critics say Parakeet beats them on accuracy and speed.
50 people gave an opinion
- recommend it
- 37
- criticize it
- 13
- approval share
- 74%
Recommended in 7 of 13 months
- November 2025: recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: not recommended
- May 2026: not recommended
- June 2026: not recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it carries the badge
- 37 people recommend itNeeds 10 or more, met.
- 2.8 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 7 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Low latency for live streaming transcription
“A 58M param model that can stream with sub-second latency on a Raspberry Pi opens up a whole class of edge applications”
Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” “The streaming architecture looks really promising for edge deployments.”
Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” “It's incredible for a live transcription stream - the latency is WOW.”
Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” - Comment from 24 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3”
- Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3”
Good accuracy for its size
“Moonshine is smaller, and I felt it was subjectively faster with about the same level of accuracy.”
Comment from 28 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” “I made moonshine the default because it has the best accuracy/latency”
Comment from 28 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” “Open-weight STT models hitting production-grade accuracy is huge”
Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” - Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3”
- Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3”
Small enough for constrained hardware
“A 58M param model that can stream with sub-second latency on a Raspberry Pi opens up a whole class of edge applications”
Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” “Moonshine is smaller, and I felt it was subjectively faster with about the same level of accuracy.”
Comment from 28 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” “small enough to actually deploy on constrained hardware”
Comment from 28 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” - Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3”
Runs locally and privately on Mac
“Transcribe any speech on your Mac completely privately.”
Comment from 26 Mar 2026 in “Private Speech Transcription for Macs – Moonshine Note Taker” “works fine on my MacOS w Tahoe”
Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3”
What people warn about
Each mark is one comment
Less accurate than some alternatives
“I tried out Moonshine and whisper and found neither of them more accurate or faster than Parakeet.”
Comment from 12 Oct 2026 in “Whistle: Speech to Text in 16.9 MB” “Moonshine was noticeably slower on my 2018-era Intel i7 PC, and didn't seem as accurate either.”
Comment from 25 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3” “looks like Parakeet V2/V3 and Canary-Qwen (a Qwen finetune) handily beat Moonshine”
Comment from 24 Feb 2026 in “Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3”
Other speech recognition models people recommend
The four most recommended other speech recognition models, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| OpenAI Whisper | 164 | 53 | 76% | 12 | — | Approved |
| NVIDIA Parakeet | 144 | 27 | 84% | 12 | — | Approved |
| Mistral Voxtral | 69 | 27 | 72% | 8 | — | Approved |
| Moonshine open-weights STT modelsThis page | 37 | 13 | 74% | 7 | — | Approved |
| whisper.cpp | 33 | 1 | 97% | 11 | 2 yrs4 owners | Approved |
What people said
40 of the 56 comments that mention it, newest first. Each links to the original comment.
“I tried out Moonshine and whisper and found neither of them more accurate or faster than Parakeet.”
“So models like moonshine and this one allow you to just embed it into your application simply.”
“You might be interested in running games from your headless boxes with this”
“starting to move over to a moonshine model”
“Streaming from my 5700xt I can get 90fps ultra. It's massively better to stream from even this old GPU than to play locally.”
“I think you're underselling this. Sunshine can't do any resolution on Linux. That's huge.”
“That does sound pretty neat”
“I played with it for 10 minutes, very impressive.”
“Moonshine creates its own compositor and does not rely on an existing desktop environment.”
“doesn't support video deckde and encoded on some older Intel ARC cards like the A770 which is a bummer”
“this option, which is new to me”
“Yes much better in my opinion.”
“This is perfect for someone who has a active desktop user and can still stream from that same PC”
“this sounds amazing for headless machines”
“seems more nimble with phonetic pronunciation in the demo”
“Moonshine is doing awesome stuff”
“super nice to be able to run it to test it like this”
“Transcribe any speech on your Mac completely privately.”
“I've found the Web Speech API pretty good, as well as moonshine”
“followed by moonshine medium”
“small enough to actually deploy on constrained hardware”
“Moonshine is smaller, and I felt it was subjectively faster with about the same level of accuracy.”
“I made moonshine the default because it has the best accuracy/latency”
“Try Moonshine with a browser GUI”
“Moonshine was noticeably slower on my 2018-era Intel i7 PC, and didn't seem as accurate either.”
“Open-weight STT models hitting production-grade accuracy is huge”
“works fine on my MacOS w Tahoe”
“the results are really great”
“I'm quite surprise to see that level of polish from an open-source project.”
“It's incredible for a live transcription stream - the latency is WOW.”
“and was surprised at the ~2% CER I was seeing in evals”
“The streaming architecture looks really promising for edge deployments.”
“Moonshine impressive too and it is half the param”
“If those numbers look good, this seems very promising for local assistant pipelines.”
“Streaming transcription is crazy fast on an M1.”
“A 58M param model that can stream with sub-second latency on a Raspberry Pi opens up a whole class of edge applications”
“looks like Parakeet V2/V3 and Canary-Qwen (a Qwen finetune) handily beat Moonshine”
“No idea why 'sudo pip install --break-system-packages moonshine-voice' is the recommended way to install on raspi?”
“The tiny streaming latencies you show look insane”
“Write a C wrapper for the best one (Moonshine)”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.