Voice2Agent adds F8 voice dictation

// 45d agoOPENSOURCE RELEASE

Voice2Agent adds F8 voice dictation

voice2agent is a lightweight push-to-talk layer for coding agents: press F8, dictate, and paste the transcript into whatever terminal, editor, or agent you’re using. It supports both local STT servers and OpenAI transcription, with a Linux/GNOME-first setup.

// ANALYSIS

This is a narrow tool, and that’s the point: it solves the annoying last mile between speech and the CLI without trying to become a full voice agent platform. The universal clipboard/paste approach is what makes it useful across Codex, Claude Code, OpenClaw, and plain terminals.

–The main value is friction reduction, not intelligence: it turns voice into a faster input method for agents you already trust
–Local HTTP STT support means privacy-sensitive users can keep audio on-device or behind an SSH tunnel
–The implementation is intentionally opinionated around Linux, GNOME shortcuts, ALSA, Wayland, and ydotool, so portability is limited
–Because it pastes text instead of typing keystrokes, it avoids keyboard-layout edge cases and works across arbitrary apps
–It sits in the same emerging category as other agent-input utilities, but differentiates by staying minimal and agent-agnostic

// TAGS

clispeechagentautomationopen-sourcevoice2agent

DISCOVERED

45d ago

2026-04-16

PUBLISHED

45d ago

2026-04-16

RELEVANCE

8/ 10

AUTHOR

thomheinrich

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

MODEL49m ago

MiniMax-M3 Hits OpenRouter with 1M Context

The MiniMax-M3 multimodal model has been added to OpenRouter, featuring a 1-million-token context window and highly competitive pricing. At just $0.30 per million input tokens and $1.20 per million output tokens, developers now have cost-effective access to state-of-the-art, long-context multimodal AI capabilities for building advanced applications and agents.

BENCHMARK53m ago

MiniMax M3 Tops GPT 5.5 on SWE-Bench Pro

The recently announced MiniMax M3 model has reportedly beaten GPT 5.5 on the SWE-Bench Pro coding benchmark, scoring 59.0% compared to GPT 5.5's 58.6%. Operating at $0.30 per million input and $1.20 per million output tokens, M3 offers immense price-to-performance potential for highly affordable agentic coding workflows.

MODEL1h ago

MiniMax launches ultra-fast M3 model

MiniMax has announced MiniMax M3, a brand new model architecture featuring a 1-million-token context window, native video input support, and up to 15.6x faster decoding speeds. The model is priced disruptively at $0.30 per million input tokens and $1.20 per million output tokens, positioning it as a highly competitive and efficient multimodal option.