Vocs: fully local macOS dictation with whisper.cpp streaming, short chunks decoded 8x faster
zenitsu--DS · reddit · 2026-10-26
Developer released Vocs, a fully on-device macOS dictation app (whisper.cpp large-v3-turbo) that pastes transcribed text into any app. MIT-licensed, built with Tauri and Rust, zero network calls in the transcription path.
Key technical details:
- Streaming chunks: audio cut at natural silence boundaries and decoded while you talk, so finish time stays flat regardless of length
- Dynamic windows: each chunk decoded with a window sized to its length instead of the full 30s, making short fragments 8x faster
- Hallucination handling: strips sound-effect tags like [BLANKAUDIO] and (laughs); repetition loops detected, retried once, then discarded
- Deterministic post-processing: Rust pass for punctuation, casing, fillers, spoken commands and a custom dictionary, table-tested to prevent silent accuracy regressions
- Optional 120MB on-device grammar model that only makes safe word-level fixes; engine sits behind an adapter interface for other local models
More from coding & agent
- How to get more ambitious with evals as models approach superhuman intelligence — shyamalanadkat · 2026-10-26
- GenAICode: open-source toolkit puts one API over Claude Code, Codex, Copilot, Cursor and Gemini CLI — Codeblix_Ltd · 2026-10-26
- Claude rebuilds a Gaussian-splatted garden scene in Blender with Geometry Nodes — CSProfKGD · 2026-10-26
- Most MCP servers skip two of four tool safety hints, breaking confirmation heuristics — abhiramreddykalluru · 2026-10-26
- 50 agent runs, zero doc reads: five models pass tests but skip the documentation — ducha_aiki · 2026-10-26
- Every's 'frontier team' model: 100 AI experiments are useless if none land — every · 2026-10-26