LOADING DATASHEET
GROUNDED IN REAL HUMAN CONVERSATIONSYOUR BENCH
parakeet tdt 0.6B
◆ OPEN
by Nvidia
OPEN WEIGHTSLOCALUS-MADE
RANKED #22 OF 32 AUDIO · OVERALL #391 OF 7,062 · VIBE SCORE 5.4 · 25 VOICES
parakeet-tdt-0.6b-v2 on Hugging Face (automatic speech recognition). 2,195,476 downloads. Open weights for local or hosted use.
HAPPY9.9
8 mentions
SAD5.1
1 mentions
VIBE TRAJECTORY
26 weeks · 37 voices
AGING WELL
The crowd is warmer now than it was at the start.
HOW TO USE IT
Open weights. You can use a hosted service, or download it and run it yourself, free.
WHAT THE INTERNET SAYS
People are mostly frustrated with parakeet tdt 0.6B right now.
24 POSTS · 1 COMMENTS · GITHUB · REDDIT · X
PEOPLE LOVE
- No fan mail yet. The crowd is still warming up.
PEOPLE COMPLAIN ABOUT
- Outages, errors, or limits get in the way (6 mentions)
TOPIC BY TOPIC
- Working when you need itA SORE SPOT
0 thumbs up · 6 thumbs down
LOOKING FOR AN ALTERNATIVE?
Other models the crowd has fully reviewed, starting with audio models like this one.
- Kokoro TTS7.3Kokoro · Audio2.9k VOICESCOMPARE
MiniMax Music 3
ElevenLabs Music> SOURCES (12)
- REDDIT · 2026-06-11 · 42 UPVOTESI built a 100% local, CPU-only voice loop for Ollama — talk to your models hands-free (Silero VAD + Parakeet STT + Supertonic TTS 3)
I built a 100% local, CPU-only voice loop for Ollama — talk to your models hands-free (Silero VAD + Parakeet STT + Supertonic TTS 3)
- REDDIT · 2025-05-23 · 40 UPVOTES🎙️ Offline Speech-to-Text with NVIDIA Parakeet-TDT 0.6B v2
🎙️ Offline Speech-to-Text with NVIDIA Parakeet-TDT 0.6B v2
- REDDIT · 2026-08-07 · 34 UPVOTESparakeet.wgsl – Fast, accurate ASR in the browser, via raw WebGPU & SIMD WASM
parakeet.wgsl – Fast, accurate ASR in the browser, via raw WebGPU & SIMD WASM
- REDDIT · 2026-06-10 · 7 UPVOTESI built a CPU-only local voice stack for AI agents (Claude Code, OpenCode, Codex) - Silero VAD + Parakeet STT + Supertonic TTS, one-command…
I built a CPU-only local voice stack for AI agents (Claude Code, OpenCode, Codex) - Silero VAD + Parakeet STT + Supertonic TTS, one-command install, macOS/Linux/Windows
- REDDIT · 2026-07-15 · 1 UPVOTESI had Grok 4.5 run an autoresearch optimization ladder on Parakeet CPU ASR. Result: Upto 2.1x faster STT
I had Grok 4.5 run an autoresearch optimization ladder on Parakeet CPU ASR. Result: Upto 2.1x faster STT
- X · 2026-08-18 · 1 UPVOTESOne of the new Personal Brains - What do you guys think?? 1. Brain — llama.cpp b10069, tuned like vLLM/sglang, GPU1 only
One of the new Personal Brains - What do you guys think?? 1. Brain — llama.cpp b10069, tuned like vLLM/sglang, GPU1 only
- GITHUB · 2026-09-22 · 0 UPVOTESwhisper: Parakeet cannot load from a symlinked HF cache (3.8.1-2)
Fixes the crashloop that #3105 shipped. Kokoro was unaffected and is running on the GPU already. What happened Parakeet downloaded correctly and then the pod crashlooped building the onnxruntime session: onnxruntime.capi.onnxruntime_pybind11_state.Fail:…
- GITHUB · 2026-09-14 · 0 UPVOTESAdd an optional Parakeet TDT v3 engine (single pass, much faster than Whisper)
Add an optional Parakeet TDT v3 engine (single pass, much faster than Whisper)
- GITHUB · 2026-09-22 · 0 UPVOTESvoice: local STT and TTS on the A2000 — Parakeet for whisper, Kokoro for TTS
Follows #3104, which built the image this uses. Ruling Tom, 2026-09-22 (#2960 / #3101): the existing RTX A2000 12 GB in talosm01 hosts vexa-whisper (3.9 GB, WHISPERTTL=-1, holds the node's only nvidia.com/gpu unit) plus Home Assistant's local STT and TTS. Not
- GITHUB · 2026-08-30 · 0 UPVOTESRun transcription on Parakeet TDT 0.6b v3 (FluidAudio), retiring whisper-server (#204)
Run transcription on Parakeet TDT 0.6b v3 (FluidAudio), retiring whisper-server (#204)
- GITHUB · 2026-09-15 · 0 UPVOTESDefault local ASR to Parakeet TDT 0.6B v2 with Whisper fallback
Default local ASR to Parakeet TDT 0.6B v2 with Whisper fallback
- GITHUB · 2026-08-21 · 0 UPVOTESAI catalog: refresh the text shortlist to Qwen3.5/Olmo 3/Granite, and trim the rows that were never the right pick
AI catalog: refresh the text shortlist to Qwen3.5/Olmo 3/Granite, and trim the rows that were never the right pick