OpenSkillIndex← back to search
claude-codeopenclaw

Openclaw-discord-voice-input-skill openclaw-discord-voice-input-skill

Install and maintain a free Discord DM voice-input bridge for an existing OpenClaw home. Use when a user wants Discord private voice messages transcribed locally with faster-whisper and forwarded into OpenClaw without paid APIs, especially on macOS launchd-based setups.

install
source · Clone the upstream repo
git clone https://github.com/TimRChen/openclaw-discord-voice-input-skill
Claude Code · Install into ~/.claude/skills/
git clone --depth=1 https://github.com/TimRChen/openclaw-discord-voice-input-skill ~/.claude/skills/timrchen-openclaw-discord-voice-input-skill-openclaw-discord-voice-input-skill
OpenClaw · Install into ~/.openclaw/skills/
git clone --depth=1 https://github.com/TimRChen/openclaw-discord-voice-input-skill ~/.openclaw/skills/timrchen-openclaw-discord-voice-input-skill-openclaw-discord-voice-input-skill
manifest: SKILL.md
source content

OpenClaw Discord DM Voice

Install the voice bridge by running 

scripts/install.py
. Prefer the script over manual patching because it copies the runtime assets, patches 
openclaw.json
, installs dependencies, renders the LaunchAgent plist, and deploys the service consistently.

Quick Start

Run:

python3 scripts/install.py --openclaw-home ~/.openclaw

Use these flags when needed:

  • --python /abs/path/to/python3
    to pin the Python interpreter used by 
    faster-whisper
  • --model base
    to set the default transcription model in the LaunchAgent
  • --language zh
    to set the default language hint
  • --skip-deploy
    to install files without starting the LaunchAgent
  • --skip-python-install
    to skip 
    pip install --user faster-whisper

What The Installer Does

  • Copy 
    assets/runtime/discord-dm-voice/*
    into 
    <openclaw-home>/runtime/discord-dm-voice
  • Install Node dependencies from the bundled 
    package-lock.json
  • Install 
    faster-whisper
    for the selected Python interpreter unless skipped
  • Set 
    tools.media.audio.enabled=false
    in 
    openclaw.json
    so OpenClaw does not race the bridge with its native audio preflight
  • Add 
    oc-voice-start
    , 
    oc-voice-restart
    , 
    oc-voice-stop
    , 
    oc-voice-status
    , and 
    oc-voice-logs
    to 
    runtime/agent-safehouse/openclaw-aliases.zsh
    when that file exists
  • Render and deploy 
    ai.openclaw.discord-dm-voice.plist

Assumptions

  • The target OpenClaw home already exists.
  • Discord is already configured in 
    openclaw.json
    , including bot token and allowlists.
  • The goal is Discord DM voice messages, not voice channels or server audio.
  • The host is macOS with 
    launchd
    .

Validate After Install

Run:

oc-voice-status
oc-voice-logs

Then send a fresh voice message to the bot in Discord DM. If the user still sees OpenClaw's old "

whisper
not in PATH" reply, re-check 
openclaw.json
and confirm 
tools.media.audio.enabled
is 
false
.

Assets

  • assets/runtime/discord-dm-voice/voice-bridge.mjs
    : Discord DM bridge
  • assets/runtime/discord-dm-voice/transcribe_with_faster_whisper.py
    : local transcription entrypoint
  • assets/runtime/discord-dm-voice/*.template
    : rendered at install time with user-specific absolute paths
  • scripts/install.py
    : canonical installer and updater