Generate TTS Cues

Pre-render the spoken announcements that downstream A/B comparison tooling concatenates with audio clips. Generated once, reused on every tune session and audition.

Resolve paths

PLUGIN_DATA_DIR="${CLAUDE_USER_DATA:-${XDG_DATA_HOME:-$HOME/.local/share}/claude-plugins}/audio-production"
TTS_DIR="$PLUGIN_DATA_DIR/tts"
VENV="$PLUGIN_DATA_DIR/venv"

Inputs

$ARGUMENTS:

• --voice=<edge-voice-id> — default en-US-AndrewNeural (clear, neutral male). See python -m edge_tts --list-voices for the catalogue.
• --engine=<edge-tts|espeak-ng> — default edge-tts. If edge-tts isn't installed or the network is unreachable, fall back to espeak-ng.
• --force — regenerate even if the cue files already exist.
• --cues="Sample 1,Sample 2,Original,Variant A,Variant B" — comma-separated list of phrases to render. Default just Sample 1,Sample 2 (sufficient for tune-preset and audition-preset).