reduce essentia code for loading the audio

[georgid]

The line calling the method stereo_to_monoChunk
https://github.com/georgid/AlignmentDuration/blob/production_no_melody/src/align/FeatureExtractor.py#L103

Could be rewrtitten in scipy like it is in separate_dsd-> train_audio:

        if (len(audioObj.shape))>1 and (audioObj.shape[1]>1):
            audioObj[:,0] = (audioObj[:,0] + audioObj[:,1]) / 2
            audioObj = audioObj[:,0]

[georgid]

solved in commit 1e58a3c