README outdated

[NovaH00]

So the example in README is outdated, you can use the following snippet for simple usage

import sounddevice as sd
import numpy as np
from openwakeword.model import Model
from openwakeword import models

oww_model = Model([models["hey_jarvis"]["model_path"]]) 

def callback(indata, frames, time, status):
    audio_frame = (indata.flatten() * 32767).astype(np.int16)
    prediction = oww_model.predict(audio_frame)
    for mdl, score in prediction.items():
        if score > 0.5:
            print(f"Detected {mdl}!")

with sd.InputStream(samplerate=16000, channels=1, blocksize=1280, callback=callback):
    while True:
        sd.sleep(1000)