WakeCore

Open hotword detection for private voice systems. Lokale Hotword-Erkennung für private KI-Systeme.

WakeCore is a cross-language SDK family for on-device hotword ("wake-word") detection. Drop a .wake file into your app, point WakeCore at it, and you get always-listening detection with no cloud calls, no API keys, no telemetry.

Languages

All languages call into the same native engine binary, which is distributed separately. See hotwords/ for sample .wake files you can use to verify your setup.

Quick start (Python)

pip install wakecore[audio]

from wakecore import Runtime, AudioCapture

with Runtime.load("hotwords/hey_computer.wake") as rt, \
     AudioCapture(rt.sample_rate, rt.frame_length) as mic:
    for frame in mic:
        if rt.process(frame):
            print("wake!")

Quick start (Node.js)

npm install wakecore

const { Runtime } = require("wakecore");

const rt = Runtime.load("hotwords/hey_computer.wake");
// feed PCM-16 mono frames of rt.bytesPerFrame bytes
rt.close();

Quick start (Rust)

# Cargo.toml
[dependencies]
wakecore = "0.1"

let mut rt = wakecore::Runtime::load("hotwords/hey_computer.wake")?;
let frame = [0i16; 512];
if rt.process(&frame)? { println!("wake!"); }

.wake files are produced by the WakeCore generator service.

License

MIT — see LICENSE.