<!-- Generated from the same content as https://heard.dev/compare/heard-vs-voicelayer. -->

# VoiceLayer vs Heard: Two Kinds of Voice Layer for Claude Code (2026)

Source: https://heard.dev/compare/heard-vs-voicelayer

What is a voice layer for Claude Code? Compare Claude Code’s /voice, VoiceLayer’s MCP tools, VoiceMode and Heard’s hook-based voice layer.

Compare · Voice layer for Claude Code · 4 min read

# What a voice layer for Claude Code is, and the two ways to build one.

“Voice layer” gets used for different things. One kind sits outside the agent and speaks for it. The other gives the agent tools to speak and listen. Heard and VoiceLayer are clear examples of each.

Claude Code /voice · VoiceLayer · VoiceMode · Heard · Checked October 6, 2026

Updated November 5, 2026

The short answer

## A voice layer lets the agent talk to you, not only you to it.

Claude Code’s `/voice` covers one direction: you speak and it transcribes your prompt. A voice layer adds the other, so the agent talks back. Two examples: Heard does it from outside the agent, and VoiceLayer from inside. Heard listens to Claude Code and Codex hooks and decides what to say, with no tools for the agent to call. [VoiceLayer](https://github.com/EtanHey/voicelayer) is an MCP server with two tools, `voice_speak` and `voice_ask`, so Claude chooses when to talk and can ask you a question out loud.

Inside means the agent controls the words and can hold a spoken back-and-forth. Outside means consistent updates in every session with no prompt changes, and one voice across several agents.

Side by side

## Four ways to add voice to Claude Code.

From Claude Code’s docs and each project’s README. Checked October 6, 2026.

- **Claude Code /voice built in:** Direction: You to Claude only; Who decides when it speaks: No speech output; Engines: Transcribed on Anthropic’s servers; Platforms: Where Claude Code runs locally, with a claude.ai account; Price: Included
- **VoiceLayer MCP server:** Direction: Both; Who decides when it speaks: The agent, by calling `voice_speak` or `voice_ask`; Engines: whisper.cpp local; edge-tts (online); Platforms: macOS, Linux; Price: Free, Apache-2.0
- **VoiceMode MCP + plugin:** Direction: Both; Who decides when it speaks: The agent, in a conversation; Engines: Local Whisper.cpp and Kokoro, or OpenAI; Platforms: macOS, Linux, Windows; Price: Free, MIT
- **Heard hooks:** Direction: Agent to you; you to agent on Power; Who decides when it speaks: Heard, from agent events; Engines: Kokoro, your own keys, or managed voices; Platforms: macOS 14+; Price: Free engine; app $30/mo (Power)

(1) Install

## How to install a voice layer for Claude Code on macOS.

To install a voice layer for Claude Code on macOS, first pick the kind you want. Heard (heard.dev), the voice context layer for agents, is a Mac app for macOS 14 or later that speaks Claude Code’s questions, approvals, results and failures from its hooks. VoiceLayer is an MCP server that gives Claude tools to speak and to listen. The steps below come from each project’s own docs and Homebrew tap, checked November 5, 2026.

Heard app: run `brew install --cask heardlabs/heard/heard` in Terminal, or [download Heard](https://heard.dev/download). Open it, sign in and enable Claude Code. The app installs its own Claude Code hook, so your next session speaks. Signing in starts 14 days of free Heard with no card; see [pricing](https://heard.dev/pricing).

Heard’s open-source engine (Apache 2.0) is built from source: `git clone https://github.com/heardlabs/heard.git && cd heard`, then `python3 -m venv .venv && source .venv/bin/activate && pip install -e .`, then `heard download-voice` (local voice, about 350 MB, one time), then `heard install claude-code`. It runs on your Mac with local Kokoro voices and needs no API key.

VoiceLayer: `brew tap etanhey/layers`, then `brew install etanhey/layers/voicelayer` for the CLI and MCP server, `brew install --cask etanhey/layers/voicebar` for the VoiceBar notch app, and `voicelayer setup`. Then add VoiceLayer to Claude Code’s `.mcp.json` as its README shows: a `voicelayer` server whose command is `socat` with args `STDIO` and `UNIX-CONNECT:/tmp/voicelayer-mcp.sock`. Give your terminal microphone access. If you only want the MCP server, install `voicelayer-mcp` from npm and add its dependencies: `brew install sox socat && pip3 install 'edge-tts>=7.3.1'`, plus `brew install whisper-cpp` and a whisper model for local speech recognition.

Terminal · Heard app on macOS

```
brew install --cask heardlabs/heard/heard
```

Terminal · VoiceLayer on macOS

```
brew tap etanhey/layers
brew install etanhey/layers/voicelayer
brew install --cask etanhey/layers/voicebar
voicelayer setup
```

(2) VoiceLayer

## VoiceLayer: speaking and listening tools for the agent.

VoiceLayer runs a persistent local daemon and exposes `voice_speak` for non-blocking speech and `voice_ask` for a spoken question that waits for your answer. It transcribes with whisper.cpp on your machine and speaks with edge-tts, Microsoft’s online voice service. On macOS it also ships VoiceBar, a small notch app. It is new and actively updated.

(3) Heard

## Heard: a voice layer outside the agent.

Heard installs hooks for Claude Code and Codex and turns their events into short spoken updates. Because it sits outside the agent, the agent spends no tokens or tool calls on talking, every session behaves the same, and updates from several sessions share one queue with the project name first. Tool-by-tool progress is not spoken. On Power you reply by pressing right ⌘, and Parrot answers questions about what your agents did.

(4) Where VoiceLayer wins

## Pick an MCP voice tool for spoken back-and-forth.

If you want Claude itself to choose its words and stop to ask you something out loud in the middle of a task, an MCP tool is the right shape. VoiceLayer is free, open source and runs on Linux as well as macOS. VoiceMode is the more established option of this kind and also runs on Windows.

(5) Where Heard wins

## Pick Heard to follow several agents without changing them.

If you run more than one session, or Codex next to Claude Code, a layer outside the agents keeps updates consistent and compressed, needs no instructions in your prompts, and keeps working when the agent forgets to call a tool.

A few more details

## Questions?

### What is a voice layer for Claude Code?

Something that lets Claude Code talk to you, not only you to it. It can be a layer outside the agent that speaks from its hooks, such as Heard, or an MCP server the agent calls, such as VoiceLayer or VoiceMode.

### Is Claude Code’s /voice a voice layer?

It is the input half. `/voice` transcribes your speech into the prompt. The terminal CLI has no built-in spoken output; users report the desktop app’s Code tab can read a single response on demand.

### Is VoiceLayer fully local?

Its speech recognition is local whisper.cpp. Its speech output uses edge-tts, which calls Microsoft Edge’s online text-to-speech service.

### Can I use VoiceLayer and Heard together?

Yes, though both may speak about the same moment. A common split is Heard for status updates and an MCP tool for spoken questions you answer.

Keep going

## Related Heard resources

[**A voice layer for coding agents** The category, explained.](https://heard.dev/voice-layer-for-coding-agents) [**Text-to-speech for Claude Code** say hook, narrator plugins, VoiceMode and Heard.](https://heard.dev/compare/claude-code-text-to-speech) [**Open-source voice tools** Which coding-agent voice tools run locally.](https://heard.dev/compare/open-source-voice-tools-for-coding-agents) [**All comparisons** Every Heard comparison page in one place.](https://heard.dev/compare)

## Give your agents a voice.

Keep the context. Get on with your work.

[Download for Mac](https://heard.dev/download/use-cases?format=dmg)

Includes 14 days of free Heard. No card needed.
