some problems

[kate123wong]

1. librosa or librasa in file:dataset preparation，On line 156 and line 159, you use librasa.

The audio data processing methods we used include `librasa`, `logfbank`, `mfcc` and `opensmile`, Here we take `librasa` 
as an example.
python dpcv/data/utils/raw_audio_process.py --mode <librasa/logfbank/opensmile> \
    --audio-dir /path/to/audio_dir \
    --output-dir /path/to/output_dir 

But I find that inraw_audio_process.py,the code in line 27 use librosa.

if self.mode == "librosa":
            self.librosa_extract(wav_file_path, video_name)

3. --video-path or --video-dir one the line 100 in the same file ,it shoud be python dpcv/data/face_extract/face_img_extractor.py --video-path /path/to/video/dir --output-dir /path/to/output/dir because face_img_extractor.py use para --video-path not --video-dir