| | |
|---|
| | | |
|---|
| | | |
|---|
| | | def build_preprocess(args, train): |
|---|
| | | if args.use_preprocessor: |
|---|
| | | if not args.use_preprocessor: |
|---|
| | | return None |
|---|
| | | if args.task_name in ["asr", "data2vec", "diar", "sv"]: |
|---|
| | | retval = CommonPreprocessor( |
|---|
| | |
|---|
| | | token_type=args.token_type, |
|---|
| | | token_list=args.token_list, |
|---|
| | | bpemodel=args.bpemodel, |
|---|
| | | non_linguistic_symbols=args.non_linguistic_symbols, |
|---|
| | | non_linguistic_symbols=args.non_linguistic_symbols if hasattr(args, "non_linguistic_symbols") else None, |
|---|
| | | text_cleaner=args.cleaner, |
|---|
| | | g2p_type=args.g2p, |
|---|
| | | split_with_space=args.split_with_space if hasattr(args, "split_with_space") else False, |
|---|
