runboubou/boubou

runboubou/boubou has 88 GitHub stars and 12 forks. Star changes are not measured yet. Last checked 2026-11-13 19:57 UTC. Last push 2026-11-13.

So werden diese Zahlen gemessen

88 Sterne 12 Forks

Gemessener Sternzuwachs

in 24h: Verlauf wird gesammelt
in 7 days: Verlauf wird gesammelt
in 15 days: Verlauf wird gesammelt
in 1 month: Verlauf wird gesammelt

Nettoänderungen in überlappenden Zeiträumen, keine Tagessummen. Fehlende Messwerte bleiben leer. Sterne zeigen Interesse, nicht Qualität.

Über uns

Run voice AI locally with open-source models. Bring it to any device with one command.

Entwicklerwerkzeuge

README.md

Original

Boubou

Talk with AI out loud, on your own machine.
Open voice models. One command. No account.

runboubou.com · Models · Phones and robots

Quick start

npx boubou@beta

Apple Silicon Mac with macOS 13.3+ and Node.js 18+. Allow microphone access, then speak.

One command. npx boubou@beta opens Boubou in your terminal.
Voices that fit your Mac. Boubou checks the chip, memory and free disk, and lists only the voices this Mac can run.
Pick a language and a voice. On runboubou.com/models, open a model, choose a language and a voice, and copy the command the page writes. Boubou sets up the rest: what hears, what thinks and what speaks. Each page says which Mac it needs.
```
npx boubou@beta run qwen3-tts:0.6b-customvoice --voice ono_anna
```
Talk from a phone or a robot. serve shares the voice: a phone scans a QR code, a robot or app takes an address and a port. Guides with pictures
```
npx boubou@beta serve qwen3-tts:0.6b-customvoice
```

Run boubou, choose a local model, speak, hear a reply, and interrupt by speaking again. An illustrative terminal workflow.

_{Illustrative conversation; setup time condensed.}
Esc pause · Enter resume · Ctrl+C quit

Linux and building from source

npx boubou@beta also opens the menu on x86-64 Linux, with the Liquid models that fit the computer. Voices that need an NVIDIA GPU are on runboubou.com/models: copy a command there and paste it into the menu's last row.

To build from source, you need Rust 1.91+ and Xcode Command Line Tools on macOS. On Linux, install a C compiler, pkg-config and ALSA development headers.

git clone https://github.com/runboubou/boubou.git
cd boubou
cargo build --locked --release
./target/release/boubou

Liquid models also run on arm64 Linux built from source. Linux needs glibc 2.38+, OpenSSL 3 and libgomp.

More commands

npx boubou@beta list
npx boubou@beta pull LFM3-Omni-2B
npx --offline boubou@beta run LFM3-Omni-2B --offline
npx boubou@beta doctor LFM3-Omni-2B
npx boubou@beta ask LFM3-Omni-2B photo.png "What is in this picture?"

serve prints a browser link and a QR code for a phone on the same Wi-Fi; one client can use the conversation at a time. Keep the printed link private: it contains the access key. --local listens only on this computer. By default, serve requests a trusted certificate from lan.runboubou.com, whose name includes the host's local IP address; its private key stays on the host. --local, --self-signed or --offline avoids that service.

Modular models also expose OpenAI-compatible Realtime and speech endpoints through serve. Prefer the printed HTTPS/WSS address; the HTTP/WS fallback is unencrypted. A robot or app that takes only an address and a port connects without the key: serve asks in its terminal before letting that address in and remembers the answer.

Each model page on the website also lists its versions and how to swap a component of a modular model. Use npx boubou@beta --help or npx boubou@beta <command> --help for every option.

Updates and the cache

The npm launcher checks for updates before a command, at most once an hour. In a terminal it asks before updating to the newest version on your release channel; in scripts it updates without asking. If the check or installation fails, it warns and uses the last available version. --offline, doctor and BOUBOU_NO_UPDATE_CHECK=1 skip the check.

Models and runtimes stay in ~/Library/Caches/boubou on macOS or ~/.cache/boubou on Linux across updates; set BOUBOU_HOME to choose another folder. A runtime update downloads only the files that changed.

How it works

Rust connects your microphone and speakers to a local model server, with echo cancellation and speech detection.

One audio model: Liquid models hear, think and speak through llama.cpp.
A modular pipeline: speech recognition → Qwen → your choice of voice. See models and runtimes.

Inference runs on your machine. No account, cloud model API key or telemetry. With run, audio stays on this computer. After downloading the models, use --offline to run without a network; with npx, pass --offline before the package name too. Update checks go through npm to your configured registry and never include audio, transcripts or model files.

Contributors

Built in the open. Thanks to everyone helping Boubou find its voice.

All contributors · Contribute to Boubou · Report a security issue

MIT.

runboubou/boubou

Gemessener Sternzuwachs

Über uns

README.md

Boubou

Quick start

How it works

Contributors

Ähnliche Projekte entdecken

codecrafters-io/build-your-own-x

public-apis/public-apis

morluto/rea

openclaw/openclaw