处理音频文件的python模块

首先是wave

def read_wav_data(filename):
	'''
	读取一个wav文件，返回声音信号的时域谱矩阵和播放时间
	'''
	filename="/home/chenyang/"+filename[2:]
	with wave.open(filename,"rb") as wav: # 打开一个wav格式的声音文件流
		num_frame = wav.getnframes() # 获取帧数
		num_channel=wav.getnchannels() # 获取声道数
		framerate=wav.getframerate() # 获取帧速率
		num_sample_width=wav.getsampwidth() # 获取实例的比特宽度，即每一帧的字节数
		str_data = wav.readframes(num_frame) # 读取全部的帧
		wav.close() # 关闭流
	wave_data = np.fromstring(str_data, dtype = np.short) # 将声音文件数据转换为数组矩阵形式
	wave_data.shape = -1, num_channel # 按照声道数将数组整形，单声道时候是一列数组，双声道时候是两列的矩阵
	wave_data = wave_data.T # 将矩阵转置
	#wave_data = wave_data 
	return wave_data, framerate  

接着就是

from python_speech_features import mfcc
from python_speech_features import delta

def GetMfccFeature(wavsignal, fs):
	# 获取输入特征
	feat_mfcc=mfcc(wavsignal[0],fs)
	feat_mfcc_d=delta(feat_mfcc,2)
	feat_mfcc_dd=delta(feat_mfcc_d,2)
	# 返回值分别是mfcc特征向量的矩阵及其一阶差分和二阶差分矩阵
	wav_feature = np.column_stack((feat_mfcc, feat_mfcc_d, feat_mfcc_dd))
	return wav_feature

接着就是

from scipy.io import wavfile as wav

fs, audio = wav.read(file)

这个是语谱图函数

def GetFrequencyFeature(wavsignal, fs):
	'''

	'''
	if(16000 != fs):
		raise ValueError('[Error] ASRT currently only supports wav audio files with a sampling rate of 16000 Hz, but this audio is ' + str(fs) + ' Hz. ')
	
	# wav波形 加时间窗以及时移10ms
	time_window = 25 # 单位ms
	window_length = fs / 1000 * time_window # 计算窗长度的公式，目前全部为400固定值
	
	wav_arr = np.array(wavsignal)
	#wav_length = len(wavsignal[0])
	wav_length = wav_arr.shape[1]
	
	range0_end = int(len(wavsignal[0])/fs*1000 - time_window) // 10 + 1 # 计算循环终止的位置，也就是最终生成的窗数
	data_input = np.zeros((range0_end, window_length // 2), dtype = np.float) # 用于存放最终的频率特征数据
	data_line = np.zeros((1, window_length), dtype = np.float)
	
	for i in range(0, range0_end):
		p_start = i * 160
		p_end = p_start + 400
		
		data_line = wav_arr[0, p_start:p_end]
		
		data_line = data_line * w # 加窗
		
		data_line = np.abs(fft(data_line)) / wav_length
		
		
		data_input[i]=data_line[0: window_length // 2] # 设置为400除以2的值（即200）是取一半数据，因为是对称的
		
	#print(data_input.shape)
	data_input = np.log(data_input + 1)
	return data_input

还用
torchaudio
https://pytorch.org/audio/
aukit