基于python语音唤醒

语音唤醒代码

没有的库自己pip

# -*- coding: utf-8 -*-
import pvporcupine
import pyaudio   # 录音库
import struct
from playsound import playsound   # 调用电脑进行播放
from TTS import audio
import time

PICOVOICE_API_KEY = "自己的去网站申请秘钥"

porcipine = pvporcupine.create(
    access_key=PICOVOICE_API_KEY,
    keyword_paths=['D:/Py_file/xiaoshuai/ppxiaoshuai.ppn'], # 自己去网站下载自己自定义的模型
)
myaudio = pyaudio.PyAudio()

stream = myaudio.open(
    input_device_index=0,
    rate=porcipine.sample_rate,
    channels=1,
    format=pyaudio.paInt16,
    input=True,
    frames_per_buffer=porcipine.frame_length
)

while True:
    audio_obj = stream.read(porcipine.frame_length, exception_on_overflow=False)
    audio_obj_unpacked = struct.unpack("h" * porcipine.frame_length, audio_obj)

    keyword_idx = porcipine.process(audio_obj_unpacked)
    if keyword_idx >= 0:
        print("识别成功")
        audio(main, "min.mp3")
        time.sleep(0.5)
        playsound("min.mp3")

TTS语音播放

简单用了一下edge_tts TTS语音库，具体你喜欢那个声音你自己去处理（去百度查下能看到音色）

import edge_tts
import asyncio

text = "主人，您找我有什么事"
voice = "zh-CN-XiaoyiNeural"  # 这个可以指定
output_file = "audio.wav"


async def my_function(text, output_file):
    tts = edge_tts.Communicate(text=text, voice=voice)
    await tts.save(output_file)


def audio(text, output_file):
    loop = asyncio.get_event_loop()
    loop.run_until_complete(my_function(text, output_file))

下面具体的扩展，你可以使用
Funasr
Whisper
Kaldi

以上来自B站大佬的学习地址如下：
https://www.bilibili.com/video/BV1Ns4y1m7TH/?spm_id_from=333.880.my_history.page.click&vd_source=c56d9b69c658f05bb1fc6f82627a29ca
https://picovoice.ai/

这个是我最新自己在训练的基于达摩院开源，也分享出来，感觉更家友好：
https://modelscope.cn/models/damo/speech_charctc_kws_phone-xiaoyun/summary