Convert log Mel bank energy to audio by your model

[zuujhyt]

Hi,
I use a toolbox from python_speech_features import logfbank to get log mel energy from wav file, can I convert the feature back to speech b waveform bby your model?
If yes, could you please guide?
Thank you.

from python_speech_features import logfbank
from scipy.io import wavfile
sample_rate, wav_data = wavfile.read('test.wav')
audio_feats = logfbank(wav_data, samplerate=sample_rate).astype(np.float32)
##
# How to convert audio_feats back to wavform?
##