From c48a1812e1eb279371b1a4fcf9a4b90cd7826582 Mon Sep 17 00:00:00 2001 From: Jonas Kantic Date: Mon, 6 Apr 2020 16:04:47 +0200 Subject: [PATCH 1/2] Renames audio files to a uniform pattern. Fixes bugs in script file. - renames noise_test/Neighbor*.wav to noise_test/NeighborSpeaking*.wav to conform to the naming in noise_train directory - renames noise_train/AirportAnnouncement_11.wav to AirportAnnouncements_11.wav to conform to the other AirportAnnouncement noise files - the config value total_snrlevels is read as int instead of float in order to avoid rutime error in np.linspace(...) - strips the list of strings which is read from the config value noise_types_excluded in order to allow comma separated lists with spaces (as suggested in the documentation of the config file) --- .../{Neighbor_1.wav => NeighborSpeaking_1.wav} | Bin .../{Neighbor_2.wav => NeighborSpeaking_2.wav} | Bin .../{Neighbor_3.wav => NeighborSpeaking_3.wav} | Bin .../{Neighbor_4.wav => NeighborSpeaking_4.wav} | Bin .../{Neighbor_5.wav => NeighborSpeaking_5.wav} | Bin .../{Neighbor_6.wav => NeighborSpeaking_6.wav} | Bin .../{Neighbor_7.wav => NeighborSpeaking_7.wav} | Bin .../{Neighbor_9.wav => NeighborSpeaking_9.wav} | Bin ...ouncement_11.wav => AirportAnnouncements_11.wav} | Bin noisyspeech_synthesizer.py | 4 ++-- 10 files changed, 2 insertions(+), 2 deletions(-) rename noise_test/{Neighbor_1.wav => NeighborSpeaking_1.wav} (100%) rename noise_test/{Neighbor_2.wav => NeighborSpeaking_2.wav} (100%) rename noise_test/{Neighbor_3.wav => NeighborSpeaking_3.wav} (100%) rename noise_test/{Neighbor_4.wav => NeighborSpeaking_4.wav} (100%) rename noise_test/{Neighbor_5.wav => NeighborSpeaking_5.wav} (100%) rename noise_test/{Neighbor_6.wav => NeighborSpeaking_6.wav} (100%) rename noise_test/{Neighbor_7.wav => NeighborSpeaking_7.wav} (100%) rename noise_test/{Neighbor_9.wav => NeighborSpeaking_9.wav} (100%) rename noise_train/{AirportAnnouncement_11.wav => AirportAnnouncements_11.wav} (100%) diff --git a/noise_test/Neighbor_1.wav b/noise_test/NeighborSpeaking_1.wav similarity index 100% rename from noise_test/Neighbor_1.wav rename to noise_test/NeighborSpeaking_1.wav diff --git a/noise_test/Neighbor_2.wav b/noise_test/NeighborSpeaking_2.wav similarity index 100% rename from noise_test/Neighbor_2.wav rename to noise_test/NeighborSpeaking_2.wav diff --git a/noise_test/Neighbor_3.wav b/noise_test/NeighborSpeaking_3.wav similarity index 100% rename from noise_test/Neighbor_3.wav rename to noise_test/NeighborSpeaking_3.wav diff --git a/noise_test/Neighbor_4.wav b/noise_test/NeighborSpeaking_4.wav similarity index 100% rename from noise_test/Neighbor_4.wav rename to noise_test/NeighborSpeaking_4.wav diff --git a/noise_test/Neighbor_5.wav b/noise_test/NeighborSpeaking_5.wav similarity index 100% rename from noise_test/Neighbor_5.wav rename to noise_test/NeighborSpeaking_5.wav diff --git a/noise_test/Neighbor_6.wav b/noise_test/NeighborSpeaking_6.wav similarity index 100% rename from noise_test/Neighbor_6.wav rename to noise_test/NeighborSpeaking_6.wav diff --git a/noise_test/Neighbor_7.wav b/noise_test/NeighborSpeaking_7.wav similarity index 100% rename from noise_test/Neighbor_7.wav rename to noise_test/NeighborSpeaking_7.wav diff --git a/noise_test/Neighbor_9.wav b/noise_test/NeighborSpeaking_9.wav similarity index 100% rename from noise_test/Neighbor_9.wav rename to noise_test/NeighborSpeaking_9.wav diff --git a/noise_train/AirportAnnouncement_11.wav b/noise_train/AirportAnnouncements_11.wav similarity index 100% rename from noise_train/AirportAnnouncement_11.wav rename to noise_train/AirportAnnouncements_11.wav diff --git a/noisyspeech_synthesizer.py b/noisyspeech_synthesizer.py index 86c136e1..031840fe 100644 --- a/noisyspeech_synthesizer.py +++ b/noisyspeech_synthesizer.py @@ -12,7 +12,7 @@ def main(cfg): snr_lower = float(cfg["snr_lower"]) snr_upper = float(cfg["snr_upper"]) - total_snrlevels = float(cfg["total_snrlevels"]) + total_snrlevels = int(cfg["total_snrlevels"]) clean_dir = os.path.join(os.path.dirname(__file__), 'clean_train') if cfg["speech_dir"]!='None': @@ -52,7 +52,7 @@ def main(cfg): filestoexclude = cfg["noise_types_excluded"].split(',') noisefilenames = glob.glob(os.path.join(noise_dir, audioformat)) for i in range(len(filestoexclude)): - noisefilenames = [fn for fn in noisefilenames if not os.path.basename(fn).startswith(filestoexclude[i])] + noisefilenames = [fn for fn in noisefilenames if not os.path.basename(fn).startswith(filestoexclude[i].strip())] filecounter = 0 num_samples = 0 From afea44e1364b950d51608b3ec5e88d71fcc92a56 Mon Sep 17 00:00:00 2001 From: Jonas Kantic Date: Mon, 6 Apr 2020 16:12:47 +0200 Subject: [PATCH 2/2] Improved stripping of values of the noise_types_excluded configuration parameter. --- noisyspeech_synthesizer.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/noisyspeech_synthesizer.py b/noisyspeech_synthesizer.py index 031840fe..ceb92b91 100644 --- a/noisyspeech_synthesizer.py +++ b/noisyspeech_synthesizer.py @@ -49,10 +49,10 @@ def main(cfg): if cfg["noise_types_excluded"]=='None': noisefilenames = glob.glob(os.path.join(noise_dir, audioformat)) else: - filestoexclude = cfg["noise_types_excluded"].split(',') + filestoexclude = [prefix.strip() for prefix in cfg["noise_types_excluded"].split(',')] noisefilenames = glob.glob(os.path.join(noise_dir, audioformat)) for i in range(len(filestoexclude)): - noisefilenames = [fn for fn in noisefilenames if not os.path.basename(fn).startswith(filestoexclude[i].strip())] + noisefilenames = [fn for fn in noisefilenames if not os.path.basename(fn).startswith(filestoexclude[i])] filecounter = 0 num_samples = 0