使用python中的.aiff文件创建深入学习模型

如何使用以前录制的.aiff文件创建深度学习模型进行培训

我正试图根据我在网上找到的一个类似的程序创建一个语音激活的目标喇叭。我几乎没有python或深度学习方面的经验，因此我的大部分代码都是从web上提取出来的，然后即兴创作，最初的概念基于所附故事中的代码

创造了自动球门喇叭的加拿大人球迷：

我已经完成了我所需要的一切，除了实际的模型（我想）。我在下面提供代码

我真的不在乎它是如何完成的，但如果不是更多的话，我希望至少有一些建议。我以前从未做过这种事

我正在Mac OS 10.13.6上使用Visual Studio代码

抱歉，下面的代码太多了。我不知道有多少是相关的

# Goal trigger custom.py
# Intended to store the last two seconds of live audio from the built-in microphone in a ring buffer, which will be fed into goalModel.is_goal()

import pyaudio
import pydub
from pydub.utils import make_chunks
import librosa
import numpy as np
import time
import requests
import sounddevice as sd
import subprocess
import os
import random
import sys
import timeit
import kbHitMod
import collections
import goalModel


# may still need to be edited to conform to my specific goals

class RingBuffer:
    """ class that implements a not-yet-full buffer """
    def __init__(self,size_max):
        self.max = size_max
        self.data = []

    class __Full:
        """ class that implements a full buffer """
        def append(self, x):
            """ Append an element overwriting the oldest one. """
            self.data[self.cur] = x
            self.cur = (self.cur+1) % self.max
        def get(self):
            """ return list of elements in correct order """
            return self.data[self.cur:]+self.data[:self.cur]

    def append(self,x):
        """append an element at the end of the buffer"""
        self.data.append(x)
        if len(self.data) == self.max:
            self.cur = 0
            # Permanently change self's class from non-full to full
            self.__class__ = self.__Full

    def get(self):
        """ Return a list of elements from the oldest to the newest. """
        return self.data




# ring buffer will keep the last 2 seconds worth of audio
ringBuffer = RingBuffer(2 * 22050)

overtime = False
print("\nOvertime mode: off\n")

def play(track_name):
    subprocess.getoutput("osascript -e 'tell application \"iTunes\" to play (first track of playlist \"Library\" whose name is \"4 pure silence\")'")

    subprocess.getoutput("osascript -e 'tell application \"iTunes\" to play (first track of playlist \"Library\" whose name is \"" + track_name + "\")'")

def callback(in_data, frame_count, time_info, flag):
    audio_data = np.frombuffer(in_data, dtype=np.float32)
    
    #audio_data = sd.rec(1,)
    # we trained on audio with a sample rate of 22050 so we need to convert it
    audio_data = librosa.resample(audio_data, 44100, 22050)
    #print(audio_data)
    ringBuffer.append(audio_data)

    state = subprocess.getoutput("osascript -e 'tell application \"iTunes\" to player state as string'")

    # machine learning model takes live audio as input and
    # decides if the last 2 seconds of audio contains a goal
    if goalModel.is_goal(ringBuffer.get()) and state == "paused":
        # GOAL!! 
        if overtime:
            play("1 New York Islanders Overtime Goal and Win Horn || NYCB Live: Home of the Nassau Veterans Memorial Coliseum")
        else:
            play("3 New York Islanders Goal Horn || NYCB Live Home of the Nassau Veterans Memorial Coliseum")
              
        # decides if the last 2 seconds of audio contains a win
    elif goalModel.is_win(ringBuffer.get()) and state != "playing":
        play("2 New York Islanders Win Horn || NYCB Live: Home of the Nassau Veterans Memorial Coliseum")

    return (in_data, pyaudio.paContinue)

pa = pyaudio.PyAudio()

stream = pa.open(format = pyaudio.paFloat32,
                 channels = 1,
                 rate = 44100,
                 output = False,
                 input = True,
                 stream_callback = callback)

# start the stream
stream.start_stream()

while stream.is_active():
    #time.sleep(0.25)
    kb = kbHitMod.KBHit()
    if kb.kbhit():    
        ot = kb.getch()
        if ot == "o":
            if overtime == False:
                overtime = True
                print("Overtime mode: ON\n")
            else:
                overtime = False
                print("Overtime mode: off\n")
        elif ot == "q":
            print("Quitting... Goodbye!\n")
            break
            

stream.close()
pa.terminate()

play("4 pure silence")
subprocess.getoutput("osascript -e 'tell application \"iTunes\" to pause'")
state = subprocess.getoutput("osascript -e 'tell application \"iTunes\" to player state as string'")
print("Program terminated. \n")

# goalModel.py
# deep learning model to detect if an Islanders goal has been scored
# by comparing real-time audio from the built-in mic to various .aiff 
# files of me reacting to goals

import numpy as np
import keras
from keras.models import Sequential
from keras.layers import Dense
import soundfile as sf
import aifc
import difflib
import os
import sys

# myFile = "YES_GOAL 1.aiff"
# aifc.open("YES_GOAL 1.aiff","r")
# nframes = aifc.getnframes()

# YES_GOAL_1 = aifc.readframes(nframes)
# print(YES_GOAL_1)
# aifc.close()

def is_goal(myFrame):
    # similarity = difflib.SequenceMatcher(None, YES_GOAL_1, myFrame)
    # if goal is detected:
    return True
    # else: 
    # return False

def is_win(myFrame):
    # if win is detected: 
    # return True
    # else:
    return False