speech_asr
2023-04-20 3e77fd44304a67a2b2253b4e56fede9762bb8464
funasr/utils/build_asr_model.py
@@ -210,7 +210,6 @@
    # frontend
    if args.input_size is None:
        # Extract features in the model
        frontend_class = frontend_choices.get_class(args.frontend)
        if args.frontend == 'wav_frontend':
            frontend = frontend_class(cmvn_file=args.cmvn_file, **args.frontend_conf)
@@ -218,7 +217,6 @@
            frontend = frontend_class(**args.frontend_conf)
        input_size = frontend.output_size()
    else:
        # Give features from data-loader
        args.frontend = None
        args.frontend_conf = {}
        frontend = None