Open swapnil3597 opened 4 years ago
"--decoder_params/infer_logits_to_pickle=True"
Have you defined "interactive_infer_params" in your configuration file? If not add that, for example:
interactive_infer_params = { "data_layer": Speech2TextDataLayer, "data_layer_params": { "num_audio_features": 64, "input_type": "logfbank", "vocab_file": "open_seq2seq/test_utils/toy_speech_data/vocab.txt", "dataset_files": [], "shuffle": False, }, }
I'm trying to transcribe the audio from pre-trained model as shown in
Streaming-ASR.ipynb
anddemo_streaming_asr.py
. I have changed theMODEL_PARAMS
inframe_asr.py
as shown below (config and checkpoint obtained from here):But when I try to initialize the FrameASR object as
asr = FrameASR()
I get the following error. How Am I suppose to providedataset_files
when I'm trying to get inference from pre-trained model.Error: