Microsoft shrinks VibeVoice-ASR to 1.58 GB for edge-CPU real-time speech recognition
Balance- · reddit · 2026-07-28
Microsoft’s VibeVoice-ASR-BitNet is a compressed variant of VibeVoice-ASR aimed at real-time inference on edge CPUs.
- Uses heterogeneous quantization to shrink the model from 4.62 GB to 1.58 GB.
- Claims 1.6–2.3× faster inference than Whisper.cpp.
- Runs with real-time capability (RTF < 1) on as few as 3 CPU threads.
- The post frames it as a practical on-device speech recognition option with no GPU required.
More from Multimodal
- 1,290-line commit lands MOSS Image support in ostris/ai-toolkit — ostrisai · 2026-10-16
- Ostris AI Toolkit adds training support for OpenMOSS's MOSS-Image-SFT model — ostrisai · 2026-10-16
- Nokia vs iPhone, car edition: AI image shows Tesla Roadster hovering over a Porsche — vaibhavbetter · 2026-10-16
- Midjourney multi-sref weight mixing recipe for cinematic light & motion portraits — michaelrabone · 2026-10-16
- MiniCPM-V 4.7 35B-A3B hits 17+ tokens/s on iPad with ~3GB memory — techNmak · 2026-10-16
- turbopixel: layer-and-mask generative compositing runs a 9B model on 4GB GPU — 3unjee · 2026-10-16