Cactus releases Whistle: a 16.9MB speech-to-text model that beats Whisper base on CPU with 6x speed
ycombinator · x · 2026-10-03
Cactus open-sourced Whistle, a speech recognition model in a single 16.9MB file that runs dependency-free on CPU, sharing the same C++ engine and quantization as their on-device foundation model Needle.
Key facts:
- Performance: 9x smaller than Whisper base, 6x faster, and mostly beats it
- Languages: English, German, French, Spanish, Italian, Dutch, Polish, with automatic language detection
- Three modes: transcription (16kHz mono, up to 30s per pass), word-level timestamps (start/end/probability), and speech embeddings (one row per 80ms frame, no decoding needed)
- Low latency: first token in 11ms; loads alongside Needle so one binary turns a clip straight into tool calls
- Target: mobiles, wearables, robots, smart home, automotive, microcontrollers — audio never leaves the device
A browser-based sandbox lets you try it directly.
More from Infra
- tinygrad lists $89,000 tinybox with 768GB VRAM: four refurbished AMD MI300X GPUs — viksit · 2026-11-19
- RustingStrata vs Strata on RTX 3060: Rust MoE engine edges out 28.3 vs 26.1 tok/s — Serious_Scarcity7818 · 2026-11-19
- Investor: market hates uncertainty more than money after Nvidia earnings — MartinGTobias · 2026-11-19
- SGLang ships Day-0 serving support for new model DaisyRose-27B — ying11231 · 2026-11-19
- Linux LTS patch flood: 3318 patches for 6.6 alone, 'no way humans are reviewing this' — pkqzy888 · 2026-11-19
- DaisyRose-27B's in-house sandbox: ~80% cheaper than cloud providers, 99.95% success rate — BanghuaZ · 2026-11-19