Acoustic Preprocssing.Transfrom Tutorial

!pip install pysensing

In this tutorial, we will be implementing a simple acoustic.preprocessing.transform

import torch
import torchaudio
import matplotlib.pyplot as plt
import librosa
import pysensing.acoustic.preprocessing.transform as transform

Load the audio

First, the example audio is loaded

# Define the plot function
def plot_wave(waveform,sample_rate):

    plt.figure(figsize=(10, 5))
    plt.plot(waveform.t().numpy())
    plt.title('Waveform')
    plt.xlabel('Sample')
    plt.ylabel('Amplitude')
    plt.tight_layout()
    plt.show()

def plot_spec(waveform,sample_rate,feature_extractor,title):
    specgram = torch.abs(feature_extractor(waveform))
    # specgram = transform.aplitude2db()(specgram)
    plt.figure(figsize=(10,5))
    plt.imshow(specgram[0].squeeze().numpy(), aspect='auto', origin='lower', cmap='inferno', extent=[0, waveform.size(1) / sample_rate, 0, sample_rate / 2])
    plt.title(title)
    plt.ylim(0,5000)
    plt.xlabel('Time (s)')
    plt.ylabel('Frequency (Hz)')

# Load the data
# Load the data
waveform, sample_rate = torchaudio.load('example_data/example_audio.wav')
plot_wave(waveform,sample_rate)

1. STFT

stft_trans = transform.stft(n_fft=2048,hop_length=1024)
plot_spec(waveform,sample_rate,stft_trans,'STFT')

2. ISTFT

stft_result = stft_trans(waveform)
ifft_trans  = transform.istft(n_fft=2048,hop_length=1024,return_complex=False)
ifft_result = ifft_trans(stft_result)
plot_wave(ifft_result,sample_rate)

3. Spectrogram

spectrogram_trans = transform.spectrogram(n_fft=2048,hop_length=1024,power=None)
plot_spec(waveform,sample_rate,spectrogram_trans,'spectrogram')

4. Inversespectrogram

spectrogram_result = spectrogram_trans(waveform)
ispec_trans  = transform.inversespectrogram(n_fft=2048,hop_length=1024)
ispec_result = ispec_trans(spectrogram_result)
plot_wave(ispec_result,sample_rate)

5. Melspectrogram

def plot_melspectrogram(specgram, title=None, ylabel="freq_bin", ax=None):
    if ax is None:
        _, ax = plt.subplots(1, 1,figsize=(10,5))
    if title is not None:
        ax.set_title(title)
    ax.set_ylabel(ylabel)
    ax.imshow(librosa.power_to_db(specgram), origin="lower", aspect="auto", interpolation="nearest")

melspectrogram_trans = transform.melspectrogram(sample_rate=sample_rate,n_fft=2048,hop_length=1024)
melspec              = melspectrogram_trans(waveform)[0].numpy()
plot_melspectrogram(melspec,'melspectrogram')

And that’s it. We’re done with our acoustic preprocessing tutorials. Thanks for reading.