Qwen Scribe Brings Qwen3-ASR Dictation to Apple Silicon in 1.2 GB of RAM
GitHub / Hacker News·low signal
A Show HN at 87 points and 18 comments ships local transcription and system-wide push-to-talk dictation for Apple Silicon Macs, at 119 stars under Apache-2.0 in v0.2.0-beta.1. It runs Qwen3-ASR through the mlx-qwen3-asr framework with a size choice between a 1.7B model for accuracy needing 3.4 GB unified memory and a 0.6B model for speed needing 1.2 GB. Features include drag-and-drop transcription with automatic language detection, right-Command push-to-talk, searchable local transcript history, and SRT export with word timestamps.