Different array size for frequency time diagram

Referring to the example at

https://de.mathworks.com/examples/wavelet/mw/deeplearning_shared-ex70694871-classify-time-series-using-wavelet-analysis-and-deep-learning

for classification of time series using wavelet analysis and deep learning where they have a uniform array size for ecg signals. In the example they have truncated the array to make it uniform.

My question is that If we have ecg signals or ECGData.Data of different length for each readings unlike that of in example what difference will it make on scalogram or on the deep neural network eventually?