diff options
| -rwxr-xr-x | datasets/dataset.pl | 4 | ||||
| -rwxr-xr-x | datasets/split44k.sh | 4 | ||||
| -rw-r--r-- | generate.py | 2 | ||||
| -rwxr-xr-x | get.pl | 22 | ||||
| -rw-r--r-- | train.py | 2 |
5 files changed, 24 insertions, 10 deletions
diff --git a/datasets/dataset.pl b/datasets/dataset.pl index 568928c..69863b7 100755 --- a/datasets/dataset.pl +++ b/datasets/dataset.pl @@ -29,7 +29,8 @@ sub process($) { } if ($ext eq 'mp3') { - system('ffmpeg', '-i', $filename, $name . '.wav'); + print "Converting mp3 to wav\n"; + system('ffmpeg', '-y', '-hide_banner', '-loglevel', 'error', '-i', $filename, $name . '.wav'); $filename = $name . '.wav'; } my $soxi = `soxi $filename`; @@ -47,7 +48,6 @@ sub process($) { } } - print "Seconds: $seconds\n"; my $a_tmp = "a_" . $filename; diff --git a/datasets/split44k.sh b/datasets/split44k.sh index b679d29..6adc2c7 100755 --- a/datasets/split44k.sh +++ b/datasets/split44k.sh @@ -11,14 +11,14 @@ dataset_path=$3 converted=".temp2.wav" rm -f $converted -ffmpeg -hide_banner -loglevel error -i $fn -ac 1 -ar 44100 $converted +ffmpeg -y -hide_banner -loglevel error -i $fn -ac 1 -ar 44100 $converted mkdir $dataset_path length=$(ffprobe -i $converted -show_entries format=duration -v quiet -of csv="p=0") end=$(echo "$length / $chunk_size - 1" | bc) echo "splitting..." for i in $(seq 0 $end); do - ffmpeg -hide_banner -loglevel error -ss $(($i * $chunk_size)) -t $chunk_size -i $converted "$dataset_path/$i.wav" + ffmpeg -y -hide_banner -loglevel error -ss $(($i * $chunk_size)) -t $chunk_size -i $converted "$dataset_path/$i.wav" done echo "done" rm -f $converted diff --git a/generate.py b/generate.py index f51b586..c56bb9a 100644 --- a/generate.py +++ b/generate.py @@ -46,7 +46,7 @@ default_params = { 'results_path': 'results', 'epoch_limit': 1000, 'resume': True, - 'sample_rate': 16000, + 'sample_rate': 44100, 'n_samples': 1, 'sample_length': 80000, 'loss_smoothing': 0.99, @@ -1,5 +1,8 @@ #!/usr/bin/perl +use DotenvSimple; +DotenvSimple::source_dotenv('.env'); + $SIG{TERM} = $SIG{INT} = sub { exit 1 }; sub sanitize ($) { @@ -21,10 +24,21 @@ my $fn, $new_fn; foreach my $s (@ARGV){ if ($s =~ /^http/) { if ($s =~ /(wav|aiff?|flac|mp3|opus)$/i) { + $s =~ s/^\s+//; + $s =~ s/\s+$//; my $fn = `basename $s`; - print "downloading $fn\n"; - system('/usr/bin/wget', $s); - system('/usr/bin/perl', 'dataset.pl', $fn); + $clean_fn = $fn; + $clean_fn =~ s/-/_/g; + $clean_fn =~ s/^\s//; + $clean_fn =~ s/\s$//; + $clean_fn =~ s/\s/_/g; + $clean_fn =~ s/_+/_/g; + system('rm', $fn); + system('rm', $clean_fn); + print "downloading $clean_fn\n"; + system($ENV{'WGET_BIN'}, '-O', $clean_fn, $s); + system('/usr/bin/perl', 'dataset.pl', $clean_fn); + $fn = $clean_fn; } else { print "youtube-dl $s\n"; my $yt = `youtube-dl --extract-audio --audio-format flac -o "%(title)s.%(ext)s" $s`; @@ -38,7 +52,7 @@ foreach my $s (@ARGV){ $new_fn = sanitize($fn); $new_fn .= '.flac'; system('mv', $fn, $new_fn); - print "got fn, $fn => $new_fn\n"; + print "youtube-dl got fn, $fn => $new_fn\n"; system('/usr/bin/perl', 'dataset.pl', $new_fn); } } @@ -46,7 +46,7 @@ default_params = { 'results_path': 'results', 'epoch_limit': 1000, 'resume': True, - 'sample_rate': 16000, + 'sample_rate': 44100, 'n_samples': 1, 'sample_length': 80000, 'loss_smoothing': 0.99, |