elevenlabs-ttslisted

# elevenlabs-tts Generate natural-sounding speech via the ElevenLabs API and hand the resulting MP3 back to the caller (typically a Telegram reply). ## Why this exists macOS `say` is functional but sounds robotic — fine for a single beep of a message, painful for anything longer than a sentence. ElevenLabs produces natural prosody that holds up over 1–2 minute messages. When a human is going to *listen* to your reply (driving, walking, eyes on something else), the quality difference matters. Prefer this skill over `say` unless the user explicitly asks for the built-in voice. ## When to fire Trigger whenever any of these are true: - A Telegram user sent a voice message and asked for an audio reply ("can you reply via audio", "I'm driving and can't read this", "send it as voice", etc.) - The user explicitly asks for audio / voice / spoken output - You're producing a spoken summary, briefing, or status update meant to be listened to rather than read - You were about to reach for macOS `say` — stop and use this instead - You're packaging a message for a human who will listen rather than read Don't fire when the user is happy with text, or when they explicitly request `say`/`afplay`/the system voice. ## How to invoke The skill is a thin wrapper around one script: ``` ~/.claude/skills/elevenlabs-tts/scripts/tts.py ``` Basic usage: ```bash python3 ~/.claude/skills/elevenlabs-tts/scripts/tts.py "Hey Coach, quick update on the booking system." ``` The script prints