tok_inference.py

source

Extract discrete speech tokens without constructing downstream losses.

usage: tok_inference.py [-h] [--config CONFIG] --output_dir OUTPUT_DIR
                        --train_config TRAIN_CONFIG --model_file MODEL_FILE
                        [--ngpu NGPU] [--dtype {float16,float32,float64}]
                        [--batch_size BATCH_SIZE] [--num_workers NUM_WORKERS]
                        [--log_level LOG_LEVEL] --data_path_and_name_and_type
                        DATA_PATH_AND_NAME_AND_TYPE [--key_file KEY_FILE]
                        [--allow_variable_data_keys ALLOW_VARIABLE_DATA_KEYS]

Named Arguments