added flag to switch ground truth visualisations between real and ide…

…alised.

visualise.py

@@ -46,6 +46,8 @@
 flags.DEFINE_boolean("closed_only", False, \
  "(Optional) Evaluate only on the closed vocabulary slice of the dataset")
 flags.DEFINE_integer("max_examples", 10, "Number of testset examples to visualise")
+flags.DEFINE_boolean("match_mels", False, \
+ "Use log() on the ground truth mel spectrogram values.")
 flags.mark_flag_as_required("pred_dataset_path")
 flags.mark_flag_as_required("ground_dataset_path")
 
@@ -71,9 +73,10 @@ def plot_mel_spectrograms(pred, y, text):
  return fig, ax
 
 def main(unused_argv):
- g_dataset_path  = FLAGS.ground_dataset_path
- p_dataset_path  = FLAGS.pred_dataset_path
+ g_dataset_path = FLAGS.ground_dataset_path
+ p_dataset_path = FLAGS.pred_dataset_path
  closed_only = FLAGS.closed_only
+ match_mels = FLAGS.match_mels
 
  # get desired book, sentence_idx
 
@@ -94,7 +97,8 @@ def main(unused_argv):
 
  g_mel_spectrogram = valid_audio_transforms(waveform).squeeze(0).transpose(0, 1)
 
- g_mel_spectrogram = torch.log(g_mel_spectrogram+1e-5)
+ if match_mels:
+ g_mel_spectrogram = torch.log(g_mel_spectrogram+1e-5)
 
  fig, ax = plot_mel_spectrograms(\
  stack_mel_spectrogram(p_mel_spectrogram),