python读取一段wav语音信号进行普通滤波、维纳滤波、卡尔曼滤波、自适应滤波等滤波方法消除或压制嗓声的代码

以下是一个Python实现的示例代码，读取WAV语音信号并应用各种滤波器进行噪声消除或压制：

import wave
import numpy as np
import matplotlib.pyplot as plt
from scipy.signal import butter, lfilter, wiener, savgol_filter
from pykalman import KalmanFilter

# 读取WAV文件
wav_file = wave.open("example.wav", "rb")

# 获取声道数、采样宽度、采样率、帧数和压缩类型
n_channels = wav_file.getnchannels()
sample_width = wav_file.getsampwidth()
sample_rate = wav_file.getframerate()
n_frames = wav_file.getnframes()

# 读取所有帧的数据
signal = wav_file.readframes(n_frames)

# 将数据转换为numpy数组
signal = np.frombuffer(signal, dtype=np.int16)

# 如果是立体声，则将两个声道的数据合并为一个
if n_channels == 2:
    signal = (signal[::2] + signal[1::2]) / 2

# 普通滤波
def butter_bandpass(lowcut, highcut, sample_rate, order=5):
    nyq = 0.5 * sample_rate
    low = lowcut / nyq
    high = highcut / nyq
    b, a = butter(order, [low, high], btype='band')
    return b, a

def butter_bandpass_filter(signal, lowcut, highcut, sample_rate, order=5):
    b, a = butter_bandpass(lowcut, highcut, sample_rate, order=order)
    filtered_signal = lfilter(b, a, signal)
    return filtered_signal

filtered_signal = butter_bandpass_filter(signal, 500, 5000, sample_rate, order=5)

# 维纳滤波
filtered_signal = wiener(signal)

# 卡尔曼滤波
kf = KalmanFilter(initial_state_mean=0, n_dim_obs=1)
filtered_signal = kf.filter(signal)[0]

# 自适应滤波
filtered_signal = np.zeros(signal.shape)
window_size = 51
for i in range(window_size//2, len(signal)-window_size//2):
    window = signal[i-window_size//2:i+window_size//2]
    mean = np.mean(window)
    std = np.std(window)
    threshold = mean + 3*std
    if signal[i] > threshold:
        filtered_signal[i] = mean
    else:
        filtered_signal[i] = signal[i]

# 绘制原始信号和滤波后的信号
plt.plot(signal, label='Original Signal')
plt.plot(filtered_signal, label='Filtered Signal')
plt.legend()
plt.show()

上述代码可以实现普通滤波、维纳滤波、卡尔曼滤波和自适应滤波四种方法来消除或压制WAV语音信号中的噪声。你可以根据自己的需求选择适合的滤波器来处理WAV语音信号数据。