From 1747f33c34e85fcbf161e4680404fdce2ee384f0 Mon Sep 17 00:00:00 2001
From: jules on spawn <spawn@cortex>
Date: Fri, 25 May 2018 15:56:35 +0200
Subject: if pitchshiftedfiles are not exactly 8 seconds remove them

---
 datasets/dataset.pl | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

(limited to 'datasets')

diff --git a/datasets/dataset.pl b/datasets/dataset.pl
index cd068bf..568928c 100755
--- a/datasets/dataset.pl
+++ b/datasets/dataset.pl
@@ -3,8 +3,8 @@
 use strict;
 use Getopt::Std;
 
-our $opt_c; # normalize the file
-our $opt_j; # jitter the pitch of each subsample by N rather than repitching big chunks
+our $opt_c; # DON'T normalize the file
+our $opt_j; # jitter the pitch of each subsample by N 1/100th semitones rather than repitching big chunks
 getopts('cj:');
 
 print "<<<< DATASET >>>>\n";
@@ -78,9 +78,16 @@ sub process($) {
       next if /^\./;
       print "$i...\n" if ++$i % 100 == 0;
       my $fn = $_;
-      my $n = rand( 2 * $opt_j ) - $opt_j;
-      system 'sox', $name . '/' . $fn, $name . '/z_' . $fn, 'pitch', $n;
-      system 'rm', $name . '/' . $fn;
+      my $in_fn = $name . '/' . $fn;
+      my $out_fn = $name . '/z_' . $fn;
+      my $n = int rand( 2 * $opt_j ) - $opt_j;
+      system 'sox', $in_fn, $out_fn, 'pitch', $n;
+      my $rep = `soxi $name/z_$fn`;
+      if ($rep =~ /352800/) {
+        unlink $in_fn;
+      } else {
+        unlink $out_fn;
+      }
     }
     closedir $dh;
   }
-- 
cgit v1.2.3-70-g09d2