speak

Speak text aloud through a Linux device's physical speaker using OpenClaw Talk.

Use this skill when the assistant should say text aloud through the device's physical speaker.

Usage

Run:

python3 {baseDir}/scripts/run.py "$SPEAK_TEXT"

where $SPEAK_TEXT is exactly the text that should be spoken.

Architecture

The skill does not call an AI provider directly and does not handle provider API credentials.

The Python wrapper:

1. calls the local OpenClaw Gateway through the openclaw CLI;
2. invokes talk.speak;
3. requests WAV output;
4. decodes the returned Base64 audio;
5. plays it through ALSA using aplay.

The active OpenClaw Talk provider owns the speech model, voice, and provider credentials.

Requirements

• Linux
• OpenClaw installed and configured
• an active OpenClaw Talk speech provider
• Python 3
• ALSA aplay

SPEAK_ALSA_DEVICE may optionally select a specific ALSA playback device.

Rules

• Pass only the text that should be spoken.
• Do not read or expose provider API keys.
• Do not call OpenAI or another speech provider directly.
• Use OpenClaw talk.speak for synthesis.