How can I solve "UnicodeDecodeError" ?
Closed this issue · 3 comments
Hello.
I am trying to train ParallelWaveGAN on my own dataset, referring to "Run training using ESPnet2-TTS recipe within 5 minutes" on this page.
However, I am getting the following error.
$ ./run.sh --stage 1 --conf conf/parallel_wavegan.v1.yaml
Stage 1: Feature extraction
Feature extraction start. See the progress via dump/eval/raw/preprocessing.*.log.
Feature extraction start. See the progress via dump/train_nodev/raw/preprocessing.*.log.
Feature extraction start. See the progress via dump/dev/raw/preprocessing.*.log.
Successfully make subsets.
Successfully make subsets.
Successfully make subsets.
run.pl: 4 / 4 failed, log is in dump/eval/raw/preprocessing.*.log
run.pl: 4 / 4 failed, log is in dump/dev/raw/preprocessing.*.log
run.pl: 4 / 4 failed, log is in dump/train_nodev/raw/preprocessing.*.log
./run.sh: 3 background jobs are failed.
The contents of preprocessing.*.log are as follows (folder and file names are changed):
# parallel-wavegan-preprocess --config conf/parallel_wavegan.v1.yaml --scp dump/train_nodev/raw/wav.1.scp --dumpdir dump/train_nodev/raw/dump.1 --verbose 1
# Started at Tue 12 Dec 2023 03:52:48 PM JST
#
0%| | 0/1174 [00:00<?, ?it/s]/mypath/anaconda3-2021.05/envs/espnet/lib/python3.8/site-packages/kaldiio/utils.py:481: UserWarning: An error happens at loading "dump/raw/org/tr_no_dev/data/format.1/wavefile.flac"
warnings.warn('An error happens at loading "{}"'.format(ark_name))
0%| | 0/1174 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/mypath/anaconda3-2021.05/envs/espnet/bin/parallel-wavegan-preprocess", line 33, in
sys.exit(load_entry_point('parallel-wavegan', 'console_scripts', 'parallel-wavegan-preprocess')())
File "/mypath/ParallelWaveGAN/parallel_wavegan/bin/preprocess.py", line 349, in main
for utt_id, (audio, fs) in tqdm(dataset):
File "/mypath/anaconda3-2021.05/envs/espnet/lib/python3.8/site-packages/tqdm/std.py", line 1178, in __iter__
for obj in iterable:
File "/mypath/ParallelWaveGAN/parallel_wavegan/datasets/scp_dataset.py", line 242, in __getitem__
fs, audio = self.audio_loader[utt_id]
File "/mypath/anaconda3-2021.05/envs/espnet/lib/python3.8/site-packages/kaldiio/utils.py", line 479, in __getitem__
return self._loader(ark_name)
File "/mypath/anaconda3-2021.05/envs/espnet/lib/python3.8/site-packages/kaldiio/matio.py", line 240, in load_mat
return _load_mat(fd, offset, slices, endian=endian)
File "/mypath/anaconda3-2021.05/envs/espnet/lib/python3.8/site-packages/kaldiio/matio.py", line 330, in _load_mat
array = read_kaldi(fd, endian)
File "/mypath/anaconda3-2021.05/envs/espnet/lib/python3.8/site-packages/kaldiio/matio.py", line 442, in read_kaldi
array = read_ascii_mat(fd)
File "/mypath/anaconda3-2021.05/envs/espnet/lib/python3.8/site-packages/kaldiio/matio.py", line 589, in read_ascii_mat
char = fd.read(1).decode(encoding=default_encoding)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x82 in position 0: invalid start byte
# Accounting: time=25 threads=1
# Ended (code 1) at Tue 12 Dec 2023 03:53:13 PM JST, elapsed time 25 seconds
It seems to be the same error as #372.
How can I solve this problem?
The simplest solution is to convert from flac to wav, and the use converted audios as the data for recipe.
Or you can use pipe command in wav.scp:
e.g., hogehoge ffmpeg -i hogehoge.flac -f wav - |
Thank you for your answer.
I will try it and report back in a few days whether it worked.
Sorry for the delay. I tried it and it worked. Thank you.