Fast, local neural text-to-speech CLI. Zero dependencies. Just speak.

Website • Releases • Installation • Quick Start

Bibo is a blazing-fast, privacy-first TTS tool that runs entirely on your machine. Built with Rust and powered by sherpa-onnx neural TTS engine.

• Zero Dependencies - No Python, no setup. TTS engine auto-downloads on first run
• Lightning Fast - Rust CLI with instant startup, 15x faster than real-time synthesis
• 100% Local - All processing on your machine, text never leaves your device
• 17 Neural Voices - Chinese, English, Korean, Russian, Spanish, German, and more
• Multi-language - MeloTTS supports Chinese + English bilingual in single voice

brew install larrykoo711/tap/bibo

Download the latest binary from Releases:

# macOS ARM64
curl -L https://github.com/larrykoo711/bibo/releases/latest/download/bibo-darwin-arm64.tar.gz | tar xz
sudo mv bibo /usr/local/bin/

# macOS x64
curl -L https://github.com/larrykoo711/bibo/releases/latest/download/bibo-darwin-x64.tar.gz | tar xz
sudo mv bibo /usr/local/bin/

# Linux x64
curl -L https://github.com/larrykoo711/bibo/releases/latest/download/bibo-linux-x64.tar.gz | tar xz
sudo mv bibo /usr/local/bin/

git clone https://github.com/larrykoo711/bibo.git
cd bibo
cargo build --release
sudo cp target/release/bibo /usr/local/bin/

# Speak text (auto-downloads default voice on first run)
bibo "Hello, world!"

# Chinese + English bilingual with MeloTTS
bibo "Hello world. 你好世界。" -v melo

# Korean
bibo "안녕하세요" -v kss

# List installed voices
bibo -l

# Show downloadable voices
bibo -d list

# Download a voice
bibo -d amy

# Save to file
bibo "Welcome" -v amy -o welcome.wav

# Fast mode (1.2x speed)
bibo -f "Quick announcement"

Run bibo -d list to see all 17 available voices.

bibo [OPTIONS] [TEXT]

Arguments:
  [TEXT]  Text to synthesize (or use -i for file, or stdin)

Options:
  -v, --voice <VOICE>    Voice to use [default: melo] [env: BIBO_VOICE]
  -s, --speed <SPEED>    Speed preset: slow, normal, fast [default: normal]
  -f, --fast             Shortcut for --speed fast
  -i, --input <FILE>     Read text from file
  -o, --output <FILE>    Save audio to WAV file
  -q, --quiet            Suppress progress output
  -l, --list             List installed voices
  -d, --download <VOICE> Download a voice model (or "list" to show all)
  -h, --help             Print help
  -V, --version          Print version

Add to ~/.claude/CLAUDE.md to let Claude speak at key moments:

Use bibo CLI for voice feedback: task done, build status, risk warnings. Keep it short.

Then Claude will say things like:

• bibo "Ship it. Tests green."
• bibo "Hold up. This looks risky."

• macOS 12+ (ARM64 / x64) or Linux (x64 / ARM64)
• ~50MB per voice model
• sherpa-onnx TTS engine (auto-downloaded on first run)

Bibo uses sherpa-onnx as the TTS engine:

1. Rust CLI - Handles arguments, file I/O, audio playback
2. sherpa-onnx - Native binary for neural TTS (VITS/MeloTTS models)

Zero Python. Zero dependencies. Everything auto-downloads on first run.

MIT License - see LICENSE for details.

• sherpa-onnx - Next-gen Kaldi TTS engine
• MeloTTS - Bilingual TTS models

• Website: larrykoo711.github.io/bibo
• Releases: github.com/larrykoo711/bibo/releases
• Homebrew: brew install larrykoo711/tap/bibo

Give voice to your AI. Ship faster.