Fakoli Speak

fakoli-speak

Multi-provider text-to-speech for Claude Code with streaming playback and cost tracking.

Supports OpenAI TTS, ElevenLabs, Deepgram Aura, Google Gemini, and macOS Say.

Quick Start

# Add your preferred provider's API key to ~/.env
OPENAI_API_KEY=sk-your-key-here         # cheapest good option ($0.015/1K chars)

# Set the provider (optional — defaults to openai)
FAKOLI_SPEAK_PROVIDER=openai

Then use /speak after any Claude response to hear it read aloud.

Commands

Command	Description
/speak	Read the last response aloud
/stop	Stop current playback
/voices	List available voices for the active provider
/status	Show playback status, provider, and config
/cost	View usage stats and spending
/provider	Show or switch the active TTS provider
/autospeak	Toggle automatic TTS on all responses

Providers

Provider	Env Var	Cost/1K chars	Quality
openai (default)	OPENAI_API_KEY	$0.015	Very good
deepgram	DEEPGRAM_API_KEY	$0.015	Good
elevenlabs	ELEVENLABS_API_KEY	$0.15–0.30	Best
google	GEMINI_API_KEY	Free tier	Good
macos	(none)	Free	Basic

Switch providers:

# Add to ~/.env to persist
FAKOLI_SPEAK_PROVIDER=openai

# Or check available providers
/provider

Provider Configuration

Each provider has its own voice and model env vars:

Variable	Default	Provider
OPENAI_TTS_VOICE	nova	openai
OPENAI_TTS_MODEL	tts-1	openai
ELEVENLABS_VOICE_ID	21m00Tcm4TlvDq8ikWAM (Rachel)	elevenlabs
ELEVENLABS_MODEL_ID	eleven_flash_v2_5	elevenlabs
DEEPGRAM_VOICE	aura-asteria-en	deepgram
GEMINI_TTS_VOICE	Kore	google
GEMINI_TTS_MODEL	gemini-2.5-flash-preview-tts	google
MACOS_SAY_VOICE	Samantha	macos

Use /voices to list available voices for the active provider.

Cost Tracking

Every TTS request logs characters, provider, and estimated cost to ~/.claude/fakoli-speak-usage.json.

/cost              # view usage summary
/cost --rate 0.015 # set your plan rate ($/1K chars)
/cost --reset      # reset usage data
/cost --json       # export as JSON

Cost rates are auto-detected per provider. Override with --rate if your plan differs.

Requirements

• Audio player — afplay (macOS, built-in), mpv, or ffplay
• Python >= 3.10
• uv for dependency management
• At least one provider API key in ~/.env (or use macos for free)

Development

make install   # install dependencies
make test      # run tests
make lint      # check syntax
make clean     # remove build artifacts

License

MIT

---

Built by Sekou Doumbouya