如果你有兩條音頻合成為一條音頻（疊加，不是拼接）的需求，以下代碼可以直接使用，需要修改的地方我已經標出來了，有三處需要修改你的本地音頻的地址：輸入音頻1，輸入音頻2，輸出音頻3。

python3.8：

#!/usr/bin/env python# -*- coding:utf-8 -*-import osimport waveimport numpy as npimport pyaudioimport librosaimport soundfile as sfimport scipy.signal as signalimport struct# ok，音頻疊加！我這里4.wav和5.wav都是5s的音頻，還沒有測試時長不同的音頻！# 參考文檔：https://www.cnblogs.com/xingshansi/p/6799994.htmlx,_ = librosa.load(’D:/4.wav’, sr=16000) #需要修改的地方：音頻1sf.write(’t1.wav’,x,16000)y,_ = librosa.load(’D:/5.wav’, sr=16000) #需要修改的地方：音頻2sf.write(’t2.wav’,y,16000)f1 = wave.open(’t1.wav’, ’rb’)f2 = wave.open(’t2.wav’, ’rb’) # 音頻1的數(shù)據(jù)params1 = f1.getparams()nchannels1, sampwidth1, framerate1, nframes1, comptype1, compname1 = params1[:6]print(nchannels1, sampwidth1, framerate1, nframes1, comptype1, compname1)f1_str_data = f1.readframes(nframes1)f1.close()f1_wave_data = np.frombuffer(f1_str_data, dtype=np.int16) # 音頻2的數(shù)據(jù)params2 = f2.getparams()nchannels2, sampwidth2, framerate2, nframes2, comptype2, compname2 = params2[:6]print(nchannels2, sampwidth2, framerate2, nframes2, comptype2, compname2)f2_str_data = f2.readframes(nframes2)f2.close()f2_wave_data = np.frombuffer(f2_str_data, dtype=np.int16) # 對不同長度的音頻用數(shù)據(jù)零對齊補位if nframes1 < nframes2: length = abs(nframes2 - nframes1) temp_array = np.zeros(length, dtype=np.int16) rf1_wave_data = np.concatenate((f1_wave_data, temp_array)) rf2_wave_data = f2_wave_dataelif nframes1 > nframes2: length = abs(nframes2 - nframes1) temp_array = np.zeros(length, dtype=np.int16) rf2_wave_data = np.concatenate((f2_wave_data, temp_array)) rf1_wave_data = f1_wave_dataelse: rf1_wave_data = f1_wave_data rf2_wave_data = f2_wave_data # ================================# 合并1和2的數(shù)據(jù)new_wave_data = rf1_wave_data + rf2_wave_datanew_wave_data = new_wave_data*1.0/(max(abs(new_wave_data)))#wave幅值歸一化new_wave = new_wave_data.tostring() p = pyaudio.PyAudio()CHANNELS = 1FORMAT = pyaudio.paInt16 # 寫文件framerate = 44100time = 10 # 產生10秒44.1kHz的100Hz - 1kHz的頻率掃描波。沒用！t = np.arange(0, time, 1.0/framerate)wave_data = signal.chirp(t, 100, time, 1000, method=’linear’) * 10000wave_data = wave_data.astype(np.short) # 打開WAV文檔f = wave.open(r'D:6.wav', 'wb') # 需要修改的地方：輸出音頻 # 配置聲道數(shù)、量化位數(shù)和取樣頻率nchannels = 1 #單通道為例sampwidth = 2data_size = len(new_wave_data)framerate = 16000 # 設置為44100就是1s，設置為8000就是10s，只有16000才是5s是對的。這里還沒搞懂！nframes = data_sizecomptype = 'NONE'compname = 'not compressed'f.setparams((nchannels, sampwidth, framerate, nframes, comptype, compname))# 將wav_data轉換為二進制數(shù)據(jù)寫入文件# f.writeframes(new_wave)for v in new_wave_data: f.writeframes(struct.pack(’h’, int(v * 64000 / 2)))f.close() # 實現(xiàn)錄音,暫時用不到。def record(re_frames, WAVE_OUTPUT_FILENAME): print('開始錄音') wf = wave.open(WAVE_OUTPUT_FILENAME, ’wb’) wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(re_frames) wf.close() print('關閉錄音')

以上就是python 實現(xiàn)音頻疊加的示例的詳細內容，更多關于python 音頻疊加的資料請關注好吧啦網其它相關文章！