语音增广（Audio Agument）

常见语音增广库

audiomentations

GitHub - iver56/audiomentations: A Python library for audio data augmentation. Inspired by albumentations. Useful for machine learning.

AugLy

GitHub - facebookresearch/AugLy: A data augmentations library for audio, image, text, and video.

nlpaug

GitHub - makcedward/nlpaug: Data augmentation for NLP

rubberband

GitHub - breakfastquay/rubberband: Official mirror of Rubber Band Library, an audio time-stretching and pitch-shifting library.

三种常用增广形式及其推荐库

音量变换

推荐：AugLy

import soundfile as sf
import augly.audio as audaugs
import numpy as np

file_path = "D:/3513-163606-0019.flac"
output_path = 'D:/new.wav'

volume_change = np.random.uniform(5,8)   #音量改变值
augmented_data, sr = audaugs.change_volume(audio=file_path, volume_db=volume_change)
sf.write(output_path, augmented_data, sr) #写入保存
     

音调变换及速度变换

推荐：rubberband（笔者下载的是rubberband命令行程序）

import subprocess
import librosa
output_file_path = 'D:/new.wav'
file_path = "D:/3513-163606-0019.flac"

rubberband_command = [
               'D:/rubberband/rubberband.exe',
               '-t', '1',  # 调整速度，1为原速
               '-p', '-1',     # 调整音调，0为原调
               file_path,
               output_file_path
           ]
# 运行 rubberband 命令
subprocess.run(rubberband_command, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
data, sr= librosa.load(file_path)
agumented_data, augmented_sr = librosa.load(output_file_path)