| | |
|---|
| | | import torch |
|---|
| | | import json |
|---|
| | | import torch.distributed as dist |
|---|
| | | import numpy as np |
|---|
| | | import kaldiio |
|---|
| | | import librosa |
|---|
| | | import torchaudio |
|---|
| | | import time |
|---|
| | | import logging |
|---|
| | | |
|---|
| | | from funasr.utils.load_utils import load_audio_text_image_video, extract_fbank |
|---|
| | | from funasr.register import tables |
|---|
| | | from funasr.utils.load_utils import extract_fbank |
|---|
| | | |
|---|
| | | |
|---|
| | | @tables.register("dataset_classes", "AudioDataset") |
|---|
| | | class AudioDataset(torch.utils.data.Dataset): |
|---|
| | |
|---|
| | | |
|---|
| | | |
|---|
| | | def collator(self, samples: list=None): |
|---|
| | | |
|---|
| | | |
|---|
| | | outputs = {} |
|---|
| | | for sample in samples: |
|---|
| | | for key in sample.keys(): |
|---|
