search

adbailey1 / DepAudioNet_reproduction

Reproduction of DepAudioNet by Ma et al. {DepAudioNet: An Efficient Deep Model for Audio based Depression Classification,(https://dl.acm.org/doi/10.1145/2988257.2988267), AVEC 2016}
66 stars 14 forks source link

How is the DAIC -WOZ Pre-processing Framework used in DepAudioNet_reproduction #7

Closed Jin-Tao-208 closed 3 years ago

Jin-Tao-208 commented 3 years ago

Hi Andrew,

As you said in the readme, using the DAIC -WOZ pre-processing framework is recommended to pre-process and extract features from the data set. But I found that the feature extraction work done by the DepAudioNet_reproduction framework overlapped with what DAIC-WOZ pre-processing did. In DAIC -WOZ pre-processing, the program extracts the participant's audio according to the time stamp, preprocesses it, and draws the mel-spectrum. However, in the framework of DepAudioNet_reproduction, it seems that there is no timestamp processing but the mel-spectrum graph is drawn directly. If so, how do I use the mel_svn_exp file generated by DAIC -WOZ pre-processing in the framework of DepAudioNet_reproduction? Hope to answer.

Thank you so much!

Best Stella

Jin-Tao-208 commented 3 years ago

Problem solved!There's something wrong with my file format.Please ignore my issue