Spectrogram augmentation
Pluto will reuse most of the wrapper functions from Chapter 7. You can reread the previous chapter if the following code seems challenging. Pluto will shorten his explanation of the wrapper functions because he assumes you are an expert at writing audio augmentation wrapper functions.
Audio Spectrogram, Mel-spectrogram, Chroma STFT, and Waveform charts take the returned amplitude data and sampling rate from the Librosa load() function reading an audio file. There is an additional transformation of the amplitude data, but they serve the same goal of visualizing the sound wave and frequencies.
After reviewing many scholarly published papers, Pluto concluded that the audio augmentation techniques in Chapter 7 apply equally well to the audio Spectrogram, Mel-spectrogram, and Chroma STFT. In particular, he referred to the scholarly paper, Audio Augmentation for Speech Recognition by Tom Ko, Vijayaditya Peddinti, Daniel Povey, and Sanjeev Khudanpur, published...
