1
$\begingroup$

I am currently trying to build an autoencoder to de-noise audio data. However I have not found any good articles explaining about the input to the autoencoder, i.e. feature vector. As in speech recognition/text up projects, MFCC is known to perform good as a input data/feature vector to neural network. I would like to know what kind of data is need for denoise autoencoder for audio/sound data.

Thank you so much for your kind help in advance.

$\endgroup$
2
$\begingroup$

As you mentioned MFCC features are one of the best features to represent audio as it captures both the time and frequency variations in the audio clip.You can get more details about MFCCS features in the below link: http://practicalcryptography.com/miscellaneous/machine-learning/guide-mel-frequency-cepstral-coefficients-mfccs/

You can import python_speech_features in python and use them to directly find the MFCCs of audio which will be in a matrix form.

You can feed the noisy audio features as input and the clean audio features as output to Deep Denoising AutoEncoder(DDAE) and then you can test on test datasets.

If required you can reconstruct the predicted mfcc output of the DDAE and check the extent to which your architecture works.

You can use Griffin Algorithm for this. Refer the link below for details about the reconstruction. https://timsainb.github.io/spectrograms-mfccs-and-inversion-in-python.html

$\endgroup$

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.