Closed takashin3391 closed 2 years ago
Could you check the file dump/raw/org/train/utt2num_samples
manually to see whether all of the values are not in the range [min_wav_duration * fs, max_wav_duration * fs]
?
@Emrys365 Thank you for your response.
The settings of enh.sh
are as follows.
fs=16k # Sampling rate. min_wav_duration=0.1 # Minimum duration in second max_wav_duration=300 # Maximum duration in second
dump/raw/org/train/utt2num_samples
is in the range [min_wav_duration fs, max_wav_duration fs] with the following values.
utt01 1766240 utt02 1766240 utt03 1766240 utt04 1766240 utt05 1766240 utt06 1766240 utt07 1766240 utt08 1766240 utt09 1766240 utt10 1766240 utt11 1766240 utt12 1766240 utt13 1766240 utt14 1766240 utt15 1766240 utt16 1766240
Does the fs notation need to be numeric? Or is 16k
fine?
By changing the value of max_wav_duration
in run.sh
in addition to max_wav_duration
in enh.sh
, all stages worked.
It is confusing to set up in multiple locations.
If possible, please improve the configuration so that it can be set in a single config file.
When 120 seconds of wav data was input to the voice enhancement task, the following error occurred in Stage 4. The error did not occur with 1 second of wav data.
The
max_wav_duration
inenh.sh
was set to 300, but the same error occurred.Is there a limit to the input length of a wav file? Where can I set the maximum length limit for input wav files?