Give Claude Code a voice. Natural text-to-speech from your terminal. 60 seconds.
Provision a tenant instantly — no email, no password, no card up front. Your API key is saved to .abz/config.json.
npm install -g abz abz signup my-app --local
Add .abz/ to your .gitignore.
Drop this block into your project's CLAUDE.md so Claude Code knows how to speak.
# tts.cx — Text-to-Speech
Turn text into natural speech. Auth: Bearer ${ABZ_API_KEY}
## Speak (CLI)
abz tts "your text" --output out.mp3
abz tts "..." --role cfo --output report.mp3 # personas: cfo cto cmo cco cso clo
abz voices # list voices, pick with -v <name>
## Speak (API)
POST https://tts.cx/api/v1/tts/synthesize
{"text":"Hello world","voice":"ava"}
→ { "audio":"<base64 mp3>", "format":"mp3", "voice":"ava" }
Ask Claude Code to voice anything. Try:
"Read this release note aloud and save it as an mp3""Generate a CFO-voiced summary of this quarter's numbers""List the available voices and pick a British one"The complete API is documented in plain text at /llms.txt, optimized for LLMs and coding agents.