Your coding agent has a voice now.

Heard speaks your coding agent's outputs so you can get up, walk around, and still know what's going on.

_{Counterpart to input tools like Wispr Flow. Wispr handles what you say to your agent; Heard handles what it says back.}

heard.dev  ·  Releases  ·  Issues

See and hear it run

Demo video coming soon. Run heard demo after install for a 15-second preview of the current voice + persona.

Install

Bring an ElevenLabs key for the best voices, or use Kokoro — free, local, no key.

Have your coding tool install it (recommended)

Paste this into Claude Code, Codex, or any AI coding tool:

Install Heard so you narrate your responses to me. Run: curl -L https://github.com/heardlabs/heard/releases/latest/download/Heard.zip -o /tmp/heard.zip && unzip -o /tmp/heard.zip -d /Applications && xattr -dr com.apple.quarantine /Applications/Heard.app && open /Applications/Heard.app — a window will pop up and I'll fill it in.

Manual

Download the latest Heard.zip, drag Heard.app into /Applications, then right-click → Open the first time (unsigned build). Onboarding walks you through voice / API key / hotkey / which agents to wire up.

CLI

pipx install heard           # or: uv tool install heard
heard install claude-code    # wires the hook
heard demo                   # ~15s preview

What it does

• Narrates tool calls + intermediate prose, not just final summaries. "Looking at your test failures." "Three failures in auth.py."
• Multi-agent aware. Run 3+ agents in parallel; Heard auto-routes narration in distinct voices so you can actually follow the work.
• Four personas, fork-your-own. Aria (calm, direct), Friday (bright, breezy), Jarvis (Marvel butler), Atlas (cinematic narrator).
• Works with any coding CLI. First-class adapters for Claude Code + Codex; heard run <command> wraps anything else.

Personas

Persona	Vibe
aria	Calm, direct, never editorial. Senior pair-programmer.
friday	Bright, breezy, three steps ahead. Sprinkles "boss".
jarvis	Marvel JARVIS-coded butler. Dry wit, "Sir" only on summaries.
atlas	Cinematic narrator. Greek tragedy applied to compile cycles.

▶ Hear the voices in action on heard.dev →

Fork your own — drop a Markdown file with frontmatter into ~/Library/Application Support/heard/personas/.

Running multiple agents

Heard auto-detects 2+ concurrent sessions and shifts mode.

Mode	When	What you hear
Solo	One session active	Full narration in your persona's voice.
Swarm	2+ sessions concurrent	Most-recent session narrates; background agents pierce only on failures and questions, with a periodic digest. Each gets a distinct voice.
Pinned	You picked one in the menu	Focus locks to that session.

Tuning

Four verbosity profiles (quiet → brief → normal → verbose), tunable globally or per-repo via .heard.yaml. Failures and wait-state questions always pierce.

Everyday commands:

heard preset jarvis              # switch persona
heard config set verbosity brief # quieter
heard silence                    # or tap Right Option to interrupt
heard doctor                     # end-to-end self-test

FAQ

Does my agent's output leave my machine?

Depends on which backends you opt into.

• Voice synth. ElevenLabs sends spoken text over HTTPS. Kokoro runs fully locally — nothing leaves the machine.
• Persona rewrites. If you provide an Anthropic key, Heard sends short candidate lines to Claude Haiku 4.5 to rewrite in-character. Skip the key and Heard uses neutral templates locally.
• Telemetry. None. No analytics, no crash reporters, no phone-home.

What does ElevenLabs actually cost in practice?

The free tier covers light daily use. A heavy day of pair-programming (2-3 hrs of narration) typically lands in the few-cents-to-low-dimes range on the paid Starter plan. Switch to Kokoro (free, local) for a hard ceiling.

Will narration slow down my agent?

No. Hooks fire-and-forget over a Unix socket; the daemon synthesises and plays asynchronously. Your agent never blocks on Heard.

Is this open source? How do I contribute?

Yes — Apache 2.0. The easiest places to contribute are adapters (heard/adapters/), personas (heard/personas/*.md), and verbosity profiles (heard/profiles/*.yaml).

Compatibility

macOS 13+ · Claude Code + Codex first-class · Cursor and Aider planned · anything else via heard run.

Status

v0.4 — multi-agent routing, profile-based verbosity, automatic ElevenLabs ⇄ Kokoro failover. Used daily by the author. APIs may still change before v1.

License

Apache 2.0.