Running distance analysis:
python run.py --layer layer_num --feat_path feat_path --output_path out_path --tsv_name tsv_name --tsv1 tsv1 --tsv2 tsv2 --group1 group1 --group2 group2 --task task_str
feat_path path of the directory with HuBERT features. Refer to the original repository for more details.
out_path path to the directory where the output csv will be created.
tsv_name name of the tsv located in the 'tsvs' directory to use. Refer to the original repository for more details.
tsv1 tsv of first group (i.e. 'ASD_parent_ran_full_processed.tsv').
tsv2 tsv of second group (i.e. 'TD_parent_ran_full_processed.tsv').
group1 name of the first group (i.e. ASD_parent; used to lookup in the dataframe).
group2 name of the second.
task_str first letter of the task to process ('L'/'O'/'C'/'N').
Preprocessing files:
python prepare_data_ran_full.py --path groups_data_path --out_path processed_groups_out_path --tsv_name tsv_name
groups_data_path path to the directory with the subdirectories of all groups ('ASD', 'ASD_parent', etc.).
processed_groups_out_path path to the directory where the processed data will be saved.
tsv_name name of the tsv the processing will create andsave in the 'tsvs' directory. Refer to the original repository for more details.
If the audio files are not in 16kHz, add --resample_waves argument. The script will convert them to 16kHz and save the converted waves with _cnv suffix.