PushToTalk is a high-performance desktop voice input tool. More than just speech-to-text, it also integrates large language model (LLM) capabilities. Hold Ctrl+Win to speak, and when released, the app automatically converts your speech to text, then polishes, translates, or formats it as an email according to your settings, and finally pastes it at the current cursor position. Supports Doubao and Qianwen.
Latest release v1.7.0 — v1.7.0 · Windows / macOS · 28 Sept 2026
3 files for Codex, OpenCode and Claude Code: push-2-talk AGENTS.md, push-2-talk CLAUDE.md, push-2-talk — 9,481 tokens loaded in every session.
AGENTS.md A 1,737 tok CLAUDE.md A 7,744 tok .claude/settings.json A — These files are yyyzl/push-2-talk's own configuration — they tell Codex, OpenCode and Claude Code how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.