Loudkit.for agents · by LoudReaderBring your agent
LOUDKIT FOR AGENTS

Your agent.
Your everyday
buddy.

Send a voice note. Hear one back.
Give the agent you already use a voice you love, in a chat that feels familiar.

Bring your agent Hear a voice

Open source · Free to run · Apple Silicon Mac

Same agent. A whole new way to talk.
A familiar voice. Your agent.01 / 04
YOUR AGENT COULD SOUND LIKE

Sophie

English · loudr-1
Voice sample

“Good morning. Take a moment to settle in, and let me tell you a story.”

Loudkit audio

Real generated audio. Press play to listen.

Real Loudkit audio. A voice sample, ready to hear.

WHEN LIFE ISN’T AT A KEYBOARD
01

On your way.

“I’ve got ten minutes. Talk me through my day.”

02

In the moment.

“Just had an idea. Help me think it through.”

03

Picking it back up.

“Where did we get to? Give me the short version.”

Example prompts. What your agent can do depends on its existing tools and permissions.

THE AGENT YOU KNOW. A VOICE YOU CHOOSE.

Less typing.
More talking.

Catch up on a walk. Think out loud. Ask the follow-up without stopping what you’re doing.

With a connected Hermes or OpenClaw bot, your agent keeps its conversation, memory and tools. Loudkit adds the voice.

  1. 01

    Say what’s on your mind.

    Send a voice note in your connected chat.

  2. 02

    Your agent does its thing.

    Your message becomes text your agent can act on.

  3. 03

    Hear a familiar voice.

    The answer comes back as audio, in a voice you’ve chosen.

FOR YOU. AND YOUR AGENT.

You pick the voice.
Your agent reads
the instructions.

Paste this into your agent’s chat. It can check your setup, explain what’s supported and help you connect the preview.

Open source. Free to run. The voice service runs on your own Mac. Get the code, choose a voice and connect your agent. Your existing agent or messenger provider may have its own costs.

Get the source on GitHub ↗
GIVE THIS TO YOUR AGENT

I'd like to send you voice notes and hear your replies with Loudkit. Read https://loudkit.loudreader.io/agents/llms.txt, check whether our agent, messenger and computer are supported, and help me install and configure the open-source voice service. Keep our existing conversation, tools and permissions, and test a voice note together.

Paste into Hermes, OpenClaw or your agent’s chat.

Open the instructions your agent will read ↗
TAKE THE INSTRUCTIONS WITH YOU

A skill for your agent.

Install the setup guide so your agent knows how to check compatibility and connect Loudkit. Both the skill and the voice service are open source and free to run on your own hardware.

Get the skill on ClawHub
Install with OpenClawnpx clawhub@0.23.3 install loudkit-voice --version 0.1.1

Run in your OpenClaw workspace. Installs the setup skill; speech models and the companion runtime are separate.

Install with Hermeshermes skills install https://loudkit.loudreader.io/.well-known/skills/loudkit-voice/SKILL.md

Installs the same setup skill from this website.

Read the skill before installing ↗Plain-text instructions for any agent ↗
FIND YOUR WAY IN

Bring the agent
you already use.

Start with a supported speech-provider setup, or explore a bridge for your own agent.

LOCAL SPEECH TESTED

Hermes & OpenClaw

Keep your existing bot and chat. Configure Loudkit for speech and Parakeet for transcription.

  • Use the agent’s existing messenger connection
  • Keep its memory, tools and permissions
  • Choose your Loudkit voice
Read the setup guide Provider tests passed. Full messenger delivery still needs testing on real accounts.
EXPLORE A CONNECTION

Your agent, your way

CLI, API and MCP are different ways in. The right one depends on what your agent exposes.

  • Codex local voice round trip tested
  • Grok Bot skill path is experimental
  • Muse Code and other adapters need runtime validation
Check compatibility A model API connection is not automatically a connection to your existing agent.
CLEAR ABOUT WHAT WORKS

Built for your chats.
Rolling out carefully.

The bridges below are implemented and protocol-tested. Live delivery on real messaging accounts has not yet been verified.

TelegramBot connection · native voice notes
DiscordBot connection · native voice notes
WhatsAppBusiness Cloud API · voice notes
SlackApp connection · audio attachments
iMessageSeparate BlueBubbles bridge · Mac required
A FEW GOOD QUESTIONS

Before we talk.

Is this a new AI assistant?

No. Loudkit supplies the voice. Your existing agent handles the conversation and its tools. The developer preview uses LoudTalk to run the local speech service and optional messenger bridges.

Does everything run offline?

Speech synthesis and transcription run on your Mac after the models download. Your messenger and your agent’s model may still use online services. This is not an offline messaging service.

Does it work with every agent?

Not yet. Hermes and OpenClaw have tested speech-provider integrations. Codex has a tested local voice pipeline. Grok Bot, Muse Code and other agents have different integration paths and remaining tests. We do not treat protocol support as proof that a complete integration works.

Can I choose the voice?

Yes. Loudkit includes 28 voices across 10 languages. Listen to the English samples above or browse the full collection. Voice choice is separate from what your agent can do.

Can I try it today?

Yes. The code is open source, with no sign-up or access request. Start with the public repository or give your agent the setup prompt above. The current voice service needs an Apple Silicon Mac. It is a developer preview: the integration guide explains which parts are tested and what still needs checking in your chat.

MAKE YOUR AGENT FEEL A LITTLE CLOSER

Your next conversation
could start with “Hey.”

Give this to your agent Find your voice

Your agent. A voice you love. A chat that feels natural.