Parakeet
NVIDIA's open-weight family of automatic speech recognition (speech-to-text) models.
HN commenters like it for fast, on-device speech-to-text1234, often running it for dictation through apps like Handy and Hex.5678 The most common complaint is that some users find it less accurate than Whisper91011, even though they agree it is faster.121314
Why people recommend it
Speed · 3
Commenters describe transcription as near-instant or a couple of seconds even on older hardware.
“near instant transcription with Parakeet V3”
“was completely surprised at just how fast it was for being on-device!”
Show 1 more
The good thing with most NNs is that you can go with fp32 almost always (I put almost because I'm not 100% sure that this covers all the cases) for almost 0 performance degradation. For Parakeet I'm 100% sure because I used it yesterday and it was using just 1GB. Why do you use GLM 5.3 tho? Unless you have an instance running "free" 24/7 you could use a local Gemma model. It should be good enough, no? Or maybe with o
Runs locally · 3
Commenters say it handles local transcription well enough to replace cloud tools for personal use.
“Nvidia's Parakeet v3 is good enough for my own just local transcription most of the time.”
“That has been my go-to model locally and it's hard to imagine there's something even better.”
Show 1 more
“Nvidia's Parakeet v3, which is the (?) leading STT model.”
Accuracy · 3
Several commenters call it reliable, and some say it beats Whisper.
“I use Parakeet V3 via this tool and it is actually quite reliable for me (in English)”
“it's perfect with parakeet.”
Show 1 more
“you can use parakeet v3 which I find way better than whisper”
Dictation app support · 3
It is commonly used inside dictation apps such as Handy, Hex, Spokenly and MacWhisper.
“It's great for me when configured to use Parakeet v3.”
“Parakeet v3 works great with it.”
Show 1 more
“Spokenly (with Parakeet as the model ) for local speech-to-text”
Voice pipelines · 3
Commenters use it as the speech-to-text stage in home voice assistants alongside local LLMs and TTS.
“I used nvidia parakeet as STT”
Show 1 more
“Parakeet for STT”
Language coverage · 2
Commenters report good results for Mandarin, and use it for English and Spanish.
“Parakeet has great mandarin>CN text”
“I use it too in Handy, for English and Spanish transcriptions”
Caveats and dissent
Accuracy vs Whisper · 3
Some commenters get better transcripts from Whisper Large, despite Parakeet being faster.
“Yes, it is much, much faster and smaller, but accuracy matters a lot”
“I keep hearing that Nvidia Parakeet models are better, but they just don't work as well for me”
Show 1 more
“I originally used Whisper and then switched to Parakeet. I much prefer whisper after playing with both.”
More errors · 1
One user says it tends to make somewhat more errors than slower models.
“mostly using parakeet v3 on my desktop, but it tends to do a bit more errors, although it is very fast”
Hard to embed · 1
One commenter says embedding it in an application is not feasible because of fragmented implementations.
“Parakeet is the gold standard.”
Punctuation commands · 1
One user says it does not respect spoken punctuation commands.
I’m a dictaphone fan. Even better than the kind shown in the article are the ones with sliding switches. I have an Olympus model. Mine has both internal memory and SD card capability. And similar to what the author aspires to, I record bits and bobs throughout the day and then use the SD card to transfer them to my Mac. From there, they automatically get added to a MacWhisper watch folder and then automatically trans
Technical jargon · 1
One commenter found a forked model handled tech jargon better than vanilla Parakeet.
“vanilla parakeet i use in handy”
Often compared with
1 moved from Whisper to Parakeet
1 moved from Parakeet to Whisper
1 moved from Parakeet to Whistle
1 moved from Parakeet to Whisper Large
Who recommends it in their own words
Threads where it came up
- Ask HN: What's the current best local/open speech-to-speech setup? 8 mentions
- Whistle: Speech to Text in 16.9 MB 7 mentions
- Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor 5 mentions
- Yap – free offline voice dictation for Mac/Windows/Linux (Wispr Flow alt) 2 mentions
- DeepSeek Introduces Vision 2 mentions
- Voxtral Transcribe 2 2 mentions
Also written as Parakeet V3, Parakeet v3, Parakeet TDT 0.6B v3, NVIDIA Parakeet. Data: parakeet.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 43
“I used nemotron 3.5 int4 and parakeet 0.6 tdt v3 int4 as the transcriber.”
“been pretty impressed with the local performance”
“ported my little voice typing gui to use Whistle/Needle instead of Parakeet”
“parakeet for STT”
The good thing with most NNs is that you can go with fp32 almost always (I put almost because I'm not 100% sure that this covers all the cases) for almost 0 performance degradation. For Parakeet I'm 100% sure because I used it yesterday and it was using just 1GB. Why do you use GLM 5.3 tho? Unless you have an instance running "free" 24/7 you could use a local Gemma model. It should be good enough, no? Or maybe with o
“Parakeet TDT 0.6B v3 (int8 CoreML, about 470 MB) runs on-device through FluidAudio's AsrManager”
“offloads transcription to Parakeet running in the browser connected to the ESP32”
“Yes, it is much, much faster and smaller, but accuracy matters a lot”
“Parakeet is the gold standard.”
“I use https://github.com/cjpais/Handy on Parakeet streaming”
“parakeet for speed”
“vanilla parakeet i use in handy”
“asr (parakeet v3), tts (omnivoice) take no more then 400-450 ms total”
“Parakeet for transcription and senior for diarization.”
“I keep hearing that Nvidia Parakeet models are better, but they just don't work as well for me”
“it's a local speech to text tool using NVIDIA Parakeet”
“(using Parakeet)”
“near realtime spoken word translation/transcription project using selfhosted Gemma4 and Nvidia Parakeet LLM models”
I’m a dictaphone fan. Even better than the kind shown in the article are the ones with sliding switches. I have an Olympus model. Mine has both internal memory and SD card capability. And similar to what the author aspires to, I record bits and bobs throughout the day and then use the SD card to transfer them to my Mac. From there, they automatically get added to a MacWhisper watch folder and then automatically trans
“near instant transcription with Parakeet V3”
“Parakeet has great mandarin>CN text”
“Nvidia's Parakeet v3 is good enough for my own just local transcription most of the time.”
“It's great for me when configured to use Parakeet v3.”
“I use Parakeet V3 via this tool and it is actually quite reliable for me (in English)”
“it's perfect with parakeet.”
“Parakeet for STT”
“This uses Parakeet + Qwen to create local transcriptions”
“Near-instant accurate-enough transcription using local Parakeet V3.”
“you can use parakeet v3 which I find way better than whisper”
“was completely surprised at just how fast it was for being on-device!”
“I use it too in Handy, for English and Spanish transcriptions”
“It utilises nvidia Parakeet as the ASR model”
“we use FluidAudio with the Parakeet-TDT 0.6b v3 model for transcription”
“benchmark it against the leading open models (Qwen3‑ASR, NVIDIA Parakeet V3, etc).”
“Spokenly (with Parakeet as the model ) for local speech-to-text”
“STT with Parakeet-V3 via Hex [1] app for near-instant good-enough transcription”
“faster transcription than Hex (with Parakeet V3)”
“I use Parakeet V3 with Handy and it works great locally for me.”
“I use the Parakeet fast model when working with Claude Code”
“stunningly fast, near-instant transcription”
“mostly using parakeet v3 on my desktop, but it tends to do a bit more errors, although it is very fast”
“Parakeet v3 works great with it.”
“I originally used Whisper and then switched to Parakeet. I much prefer whisper after playing with both.”