Open source

Compare · Heard vs. Aloud · 5 min read

Heard vs. Aloud: per-session alerts, or one voice for every session.

Aloud and Heard both listen to Claude Code and Codex hooks on a Mac and speak questions, approvals and failures before anything else. The differences are in how you switch them on, what they say about the rest, and what comes with them.

From Aloud’s README and Heard’s docs · Checked October 21, 2026

The short answer

Pick Aloud for free, local speech in the sessions you choose; pick Heard to follow every session at once.

Aloud is a free, MIT-licensed macOS tool by softcane that reads Claude Code and Codex alerts aloud. Type aloud on inside a session and it speaks questions, plan approvals, permission requests and failures first, then a short gist of each finished reply. Speech runs on your Mac with Kokoro, and sessions you have not switched on stay silent.

Heard (heard.dev), the voice layer for coding agents, also hooks into Claude Code and Codex on macOS, but it speaks for every session once you enable the agent, lets you pick how much you hear with listening modes, and in the app batches routine updates from several sessions into short summaries that start with the project name. Its open-source engine is free; the Mac app has paid plans with managed voices, spoken recaps and, on Power, voice replies. Aloud wins on price, on per-session control, and on local speech with no account from a pipx install; Heard’s local option means building its open-source engine from source.

Side by side

What each one does.

Aloud facts from its README. Checked October 21, 2026.

ToolWhat you hearSwitching it onVoiceAgentsReply by voicePrice
AloudsoftcaneQuestions, approvals and failures, then a gist of each reply; aloud full for the whole replyPer session: type aloud onKokoro on your Mac, American EnglishClaude Code, CodexNot in its READMEFree, MIT
Heardvoice layerQuestions, approvals and failures in every mode; results and routine steps by mode (Focus keeps both quiet); several sessions batched by project in the appPer agent, for all its sessions; listening modes set how muchLocal Kokoro or your own keys (engine); managed voices (app)Claude Code, Codex CLI and Codex app; status for CursorHold right ⌘ (Power)Engine free; app Pro $15/mo, Power $30/mo, 14-day Power trial

(1)Aloud

What Aloud does well.

Aloud is small and focused. Its hooks catch questions, plan approvals, permission requests, elicitation, failures and finished replies, and it also checks the transcripts of switched-on sessions once a second for events that only show up there. Questions come first, a higher-priority alert can interrupt a lower one, and a new prompt stops stale playback. Before speaking, it strips Markdown, URLs and code bodies and redacts common credential patterns.

Everything runs on your Mac: Kokoro stays loaded in a background daemon, and on first use the Kokoro package may download its model and voice files from Hugging Face. There is no account. aloud repeat replays the last alert, aloud full reads the whole last reply, and keyboard shortcuts through Hammerspoon do the same without typing.

  • Free and open source (MIT).
  • Local speech, no account.
  • Silent until you switch a session on.
  • Full reply on request.
Terminal · Aloud
brew install [email protected] pipx
pipx ensurepath
pipx install --python python3.11 git+https://github.com/softcane/aloud.git
aloud install
aloud doctor

(2)Heard

What Heard does differently.

Heard is built for following several agents without watching any of them. Once you enable Claude Code or Codex in the app, every session can speak, and Heard decides what deserves speech: routine tool calls get a one-liner or stay quiet, depending on the listening mode (Co-pilot, Companion or Focus). Questions, approvals and failures always come through as a sentence or two, and results do too in Co-pilot and Companion; Focus keeps routine results quiet. In the app, with several sessions running, routine updates are batched into short summaries that start with the project name.

The app adds managed voices and spoken recaps on Pro. On Power you hold right ⌘ to dictate a reply into the right session, ask Parrot questions about a repo, and follow along on the iPhone companion. The open-source engine, Apache 2.0, is a separate build you run yourself with local Kokoro voices.

  • Every session, no per-session command.
  • Summaries across sessions, project name first.
  • Listening modes.
  • Voice replies and iPhone on Power.

(3)Where Aloud wins

Pick Aloud for speech in the few sessions you choose.

If you want speech in one or two sessions and silence everywhere else, Aloud’s aloud on is exactly that. It costs nothing, needs no account, and keeps the text being spoken on your Mac. It can also read a whole reply when you ask with aloud full.

(4)Where Heard wins

Pick Heard to follow many sessions, and to answer back.

If you run several Claude Code or Codex sessions and do not want to arm each one, Heard covers them all and keeps them apart by project. Listening modes let you go quiet without switching anything off, and on Power you answer an agent’s question by voice without going back to its window.

(5)Both

Running them together.

Both install Claude Code and Codex hooks, so in a session where Aloud is on you would hear two voices about the same moment. Pick one tool per agent: enable Claude Code or Codex in Heard, or turn Aloud on in that agent’s sessions, not both. If you need everything to stay on your Mac, the choice is Aloud or Heard’s open-source engine with Kokoro.

Questions

Frequently asked questions

Is Aloud free?

Yes. Aloud is open source under the MIT license and installs with pipx. It needs macOS, Python 3.11 or 3.12, Homebrew, and Claude Code or Codex.

Does Aloud run locally?

Yes. Aloud speaks with Kokoro on your Mac and sends no reply text to a hosted text-to-speech service. On first use the Kokoro package may download its model and voice files from Hugging Face.

Does Aloud read every reply in full?

No. It speaks alerts and a short gist, about 240 characters, of each finished reply. aloud full or Cmd+Ctrl+H reads the whole last reply.

What is the main difference between Heard and Aloud?

Aloud speaks only in sessions where you type aloud on, with a local voice, for free. Heard (heard.dev), the voice layer for coding agents, speaks for every session of the agents you enable, batches routine updates from several sessions by project in the app, and on Power lets you reply by voice.

Do they work outside macOS?

No. Aloud requires macOS, and Heard runs on macOS 14 or later.

Ready when your agents are

Hear the work without watching every screen.

Start with the full Heard Power experience for 14 days. No card required. The open source engine remains free.

Download Heard for Mac macOS 14 or later · Claude Code and Codex supported