How to get an audio file from a power spectogram image?

Question

Python

Good afternoon. I am converting audio to picture with the following code without converting stft to absolute numbers so as not to lose data. As I understand it, I don’t lose anything at all, in order to restore the original sound, correct me if it’s wrong.

path = '/content/drive/MyDrive/originalputin/2.mp3.wav'

audio, sr = librosa.load(path)

print('length of the audio:',len(audio)/sr)

audio_stft = librosa.stft(audio)

audio_stft_db = librosa.power_to_db(audio_stft)

print('audio_stft.shape:',audio_stft.shape)
print('audio_stft_db.shape:',audio_stft_db.shape)

import matplotlib.pyplot as plt
import librosa.display

fig, ax = plt.subplots()
img = librosa.display.specshow(audio_stft_db, y_axis='log', x_axis='time', ax=ax)
ax.set_title('Power spectrogram')
fig.colorbar(img, ax=ax, format="%+2.0f dB")

At the output I get a picture. Something like this.



Questions.

1) How can I download only the picture of this spectogram without losing quality? To without frames, without a scale. Most importantly, no loss of quality at all.

2) How can I get the original audio file again from this picture without losing quality?

Thank you.