I am trying to develop a mispronunciation detection model for English speech. I use TIMIT dataset, this is phoneme labeled audio dataset.

A phoneme is any of the perceptually distinct units of sound. So, my dataset looks like an audio file and string of phonemes corresponding to that audio. Ex:

SX141.wav -> p l eh zh tcl t ax-h pcl p axr tcl t ih s pcl p ey dx ih n ax v aa dx ix z ix kcl k w aa dx ix kcl k ah m pcl p tcl t ih sh ix n


So, the problem is overfitting. My model is very good at training, but poor on testing. So because of this, I want to try synthetically increase my dataset. Maybe change the speed of audio or add some background noises etc.

Are there any already-ready solutions for doubling the audio dataset? Or, how to change speed and add some noises on the audio file? Will be it helpful?


I did not find the ready solution for it. I solve this task by myself.

  1. Increase speed.

     from scipy.io.wavfile import read, write
     Fs, data = read(filename)
     write(destination, int(Fs*1.25), data)

I save the file and increase its frequency by 1.25.

  1. Add noise.

     import numpy as np
     from scipy.io.wavfile import read, write
     Fs, data = read(filename)
     data_noise = np.random.normal(0, .2, data.shape)
     write(destination, int(Fs), data+data_noise) 

Here I generate the noise array and add it to the original wav signal.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.