voice_bot_prototype.py

import pyaudio
import wave
from array import array
import subprocess
import numpy
import soundfile as sf
import io
import base64
import requests
import json
from AI_ESB import ai
import subprocess

FORMAT = pyaudio.paInt16
CHANNELS = 1
RATE = 16000
CHUNK = int(16000/2)
RECORD_SECONDS = 15
THRESHOLD = 5000


def main():
    # recording prerequisites
    talk()
    # end of recording


def google_stt(audio):
    g = base64.b64encode(audio)
    data = '{"config": { "encoding":"FLAC","sampleRateHertz":16000,"languageCode":"cmn-Hant-TW"},"audio": { "content": "%s" }}' % g.decode(
        'utf-8')
    url = "https://speech.googleapis.com/v1/speech:recognize"
    querystring = {"key": "要填自己的喔"}
    headers = {'cache-control': "no-cache"}
    response = requests.request(
        "POST", url, data=data, headers=headers, params=querystring)
    ans = json.loads(response.text)
    text = ''
    if (ans == {}):
        return text
    for alt in ans['results']:
        for item in alt['alternatives']:
            text = text + item['transcript']
    print(text)
    return text


def t2s(text):
    file = "output2.mp3"
    url = "https://translate.google.com/translate_tts"
    querystring = {"ie": "UTF-8", "total": "1", "idx": "0",
                   "textlen": "128", "client": "tw-ob", "q": text, "tl": "zh-TW"}
    headers = {'cache-control': "no-cache"}
    response = requests.request(
        "GET", url, headers=headers, params=querystring, stream=True)
    with open(file, 'wb') as f:
        for chunk in response.iter_content(chunk_size=1024):
            if chunk:  # filter out keep-alive new chunks
                f.write(chunk)
                f.flush()
    #subprocess.call('/usr/bin/play %s' % file, shell=True)
    subprocess.call(['/usr/bin/afplay', file])


def talk():
    # starting recording
    while 1:
        # try:
        audio = pyaudio.PyAudio()  # instantiate the pyaudio
        stream = audio.open(format=FORMAT, channels=CHANNELS,
                            rate=RATE,
                            input=True,
                            frames_per_buffer=CHUNK)
        frames = []
        status = 'ready'
        while 1:
            data = stream.read(CHUNK)
            data_chunk = array('h', data)
            vol = max(data_chunk)
            print(vol)
            if(vol >= THRESHOLD and status == 'ready'):
                print("something said")
                frames.append(data)
                status = 'recording'
                print(status)
            elif(status == 'ready'):
                frames = [data]
            elif(vol >= THRESHOLD and status == 'recording'):
                frames.append(data)
            elif(vol < THRESHOLD and status == 'recording'):
                status = 'stop'
                frames.append(data)
                print(status)
                break
            else:
                print("nothing")
        print('save...\n')
        # 取出wave buffer轉到 numpy
        decoded = numpy.fromstring(b''.join(frames), 'Int16')
        flac_buffer = io.BytesIO()
        # 將numpy 轉成 flac binary
        sf.write(flac_buffer, decoded, samplerate=RATE,
                 format="FLAC", subtype="PCM_16")
        print('save...\n')
        google_text = google_stt(flac_buffer.getvalue())
        if (google_text != ''):
            pass
            t = ai.talk(google_text)
            # print(t)
            t2s(t)
        else:
            print("QQ沒有回應")
        stream.stop_stream()
        stream.close()
        audio.terminate()


        # except :
        #  print('terminal...\n')
        #  break
if __name__ == "__main__":
    main()