Is there any reference to use CWT layer for audio dataset

Is there any reference to use CWT layer for audio dataset

https://in.mathworks.com/help/wavelet/ug/time-frequency-convolutional-network-for-eeg-data-classification.html

I am refering the above code but the they used txt fie dataset where as i have to use audio dataset with raw waveform modelling technique using CWT layer as 1st layer of the network.

Answer 1

[Hassaan]

To adapt the code for your audio dataset with a CWT layer as the first layer, you can follow these general steps:

1. Data Preparation: Load and preprocess your audio dataset. This will involve reading the audio files, extracting features or converting the waveform data into a format suitable for input to a neural network.
2. Network Architecture: Modify the network architecture to accommodate audio data. Replace the input layer with a CWT layer as the first layer. You will need to configure the CWT layer with appropriate parameters such as wavelet type, scales, and so on.
3. Training: Train the modified network using your audio dataset. Make sure to adjust the network's input size and output size to match your classification task.
4. Evaluation: Evaluate the trained network's performance on your audio dataset.

Here's a high-level example of how you can modify the code:

% Load and preprocess your audio dataset
% Replace this part with code to load and preprocess audio data

% Define the CWT layer as the first layer
waveletName = 'cmor1.5-1';
scales = 1:64; % Adjust the scales as needed
cwtLayer = cwtfilterbank('Wavelet', waveletName, 'SignalLength', signalLength, 'VoicesPerOctave', 12, 'Scales', scales);

% Modify the network architecture
layers = [
    imageInputLayer([1 signalLength 1]) % Adjust the input size
    cwtLayer
    convolution2dLayer([1 3], 16, 'Padding', 'same')
    % Add more layers as needed for your task
    fullyConnectedLayer(outputSize) % Adjust outputSize for your task
    softmaxLayer
    classificationLayer
];

% Train the network
% Replace this part with code to train the network using your audio dataset

% Evaluate the network
% Replace this part with code to evaluate the network on your audio dataset

Please note that you will need to adapt the code further to match your specific audio dataset and classification task. Additionally, you may need to adjust hyperparameters, the network architecture, and other settings based on your dataset's characteristics. The key modification is replacing the input layer with a CWT layer as the first layer of the network.

---------------------------------------------------------------------------------------------------------------------------------------------------

If you find the solution helpful and it resolves your issue, it would be greatly appreciated if you could accept the answer. Also, leaving an upvote and a comment are also wonderful ways to provide feedback.

Professional Interests

• Technical Services and Consulting
• Embedded Systems | Firmware Developement | Simulations
• Electrical and Electronics Engineering

Feel free to contact me.

Answer 2

[Sajiha Shaik]

But i am getting another error as

Error using nnet.cnn.layer.ImageInputLayer/vertcat

Cannot convert from 'cwtfilterbank' to 'nnet.cnn.layer.ImageInputLayer'.