Soohwan Kim comments

Results 80 comments of


                                            Soohwan Kim

Questions about the format and shape of the data

1. Spectrogram 2. Ignore everything behind .

Questions about the format and shape of the data

Sorry for the late response. I recommend checking [this project](https://github.com/openspeech-team/openspeech)

Questions about the format and shape of the data

Show me the code.

Questions about the format and shape of the data

@zwan074 Check [this link](https://github.com/openspeech-team/openspeech/blob/14ae8727c9f547c75ec572a0e563aaf5aedbd1fa/openspeech/models/openspeech_encoder_decoder_model.py#L111)

Questions about the format and shape of the data

@jcgeo9 289 is almost a quarter of 1162. This phenomenon occurs due to [Conv2dSubampling](https://github.com/sooftware/conformer/blob/aead2f267157726b088eb301207a64aa983b6cc2/conformer/convolution.py#L152) during the convolution block of the Conformer.

Questions about the format and shape of the data

I updated the code and README because many people seemed to have a hard time calculating losses. Below is an example of calculating CTC Loss. ```python import torch import torch.nn...

the Usage sample got an invalid size error

Did you use this code? ```python import torch import torch.nn as nn from conformer import Conformer batch_size, sequence_length, dim = 3, 12345, 80 cuda = torch.cuda.is_available() device = torch.device('cuda' if...

the Usage sample got an invalid size error

Do you want to try it without nn.DataParallel?

the Usage sample got an invalid size error

I'm operating normally. It's weird.

The meaning of Usage

Check https://github.com/openspeech-team/openspeech