Librispeech v2

bootphon · Mar 20, 2015 · ef71e1f · ef71e1f
1 parent 1266eda
commit ef71e1f
Show file tree

Hide file tree

Showing 5 changed files with 56 additions and 11 deletions.
diff --git a/corpora/LibriSpeech_train-clean-100/bin/copy_phones.sh b/corpora/LibriSpeech_train-clean-100/bin/copy_phones.sh
@@ -1,2 +1,3 @@
 #copy phones.txt from if doesn't already exist
-cp -u /fhgfs/bootphon/data/raw_data/LibriSpeech/phones.txt /fhgfs/bootphon/scratch/xcao/github_abkhazia/abkhazia/corpora/LibriSpeech/data/
+cd /fhgfs/bootphon/scratch/xcao/github_abkhazia/abkhazia/corpora/LibriSpeech_train-clean-100/data/
+cp -f -u /fhgfs/bootphon/data/raw_data/LibriSpeech/phones.txt ./phones.txt
diff --git a/corpora/LibriSpeech_train-clean-100/bin/data_prep.py b/corpora/LibriSpeech_train-clean-100/bin/data_prep.py
@@ -25,7 +25,7 @@
 # paths - needs to change paths and versions of Librispeech
 # create 'data' directory if doesn't exist
 raw_path = "/fhgfs/bootphon/data/raw_data/LibriSpeech/"
-dict_path = "/fhgfs/bootphon/data/raw_data/cmu_combined/"
+dict_path = "/fhgfs/bootphon/data/raw_data/CMU_dict/"
 derived_path = "/fhgfs/bootphon/data/derived_data/LibriSpeech_abkhazia/data/"
 github_path = "/fhgfs/bootphon/scratch/xcao/github_abkhazia/abkhazia/corpora/LibriSpeech_train-clean-100/data/"
 
@@ -138,6 +138,8 @@ def text(trs, wav):
 	outfile = open(output_file_text, "w")
 	outfile2 = open(output_corrupted_wavs, "w")
 	input_dir = os.path.join(derived_path, trs)
+	if not os.path.isdir(input_dir):
+		os.makedirs(input_dir)
 	input_dir_wav = os.path.join(derived_path, wav)
 	files = os.listdir(input_dir)
 	wav_list = os.listdir(input_dir_wav)

diff --git a/corpora/LibriSpeech_train-clean-360/bin/copy_phones.sh b/corpora/LibriSpeech_train-clean-360/bin/copy_phones.sh
@@ -1,2 +1,3 @@
 #copy phones.txt from if doesn't already exist
-cp -u /fhgfs/bootphon/data/raw_data/LibriSpeech/phones.txt /fhgfs/bootphon/scratch/xcao/github_abkhazia/abkhazia/corpora/LibriSpeech/data/
+cd /fhgfs/bootphon/scratch/xcao/github_abkhazia/abkhazia/corpora/LibriSpeech_train-clean-360/data/
+cp -f -u /fhgfs/bootphon/data/raw_data/LibriSpeech/phones.txt ./phones.txt
diff --git a/corpora/LibriSpeech_train-clean-360/bin/data_prep.py b/corpora/LibriSpeech_train-clean-360/bin/data_prep.py
@@ -25,7 +25,7 @@
 # paths - needs to change paths and versions of Librispeech
 # create 'data' directory if doesn't exist
 raw_path = "/fhgfs/bootphon/data/raw_data/LibriSpeech/"
-dict_path = "/fhgfs/bootphon/data/raw_data/cmu_combined/"
+dict_path = "/fhgfs/bootphon/data/raw_data/CMU_dict/"
 derived_path = "/fhgfs/bootphon/data/derived_data/LibriSpeech_abkhazia/data/"
 github_path = "/fhgfs/bootphon/scratch/xcao/github_abkhazia/abkhazia/corpora/LibriSpeech_train-clean-360/data/"
 
@@ -138,6 +138,8 @@ def text(trs, wav):
 	outfile = open(output_file_text, "w")
 	outfile2 = open(output_corrupted_wavs, "w")
 	input_dir = os.path.join(derived_path, trs)
+	if not os.path.isdir(input_dir): 
+		os.makedirs(input_dir)
 	input_dir_wav = os.path.join(derived_path, wav)
 	files = os.listdir(input_dir)
 	wav_list = os.listdir(input_dir_wav)
@@ -247,15 +249,15 @@ def copy_phones():
 
 
 #Running the different steps
-copy_flac ('train-clean-360','flac_train-clean-360')
-convert_flac()
-copy_trs ('train-clean-360','trs_train-clean-360')
-convert_speaker_ID_wav ('wav_train-clean-360')
-link_wavs()
-segments_speakers('wav_train-clean-360')
+#copy_flac ('train-clean-360','flac_train-clean-360')
+#convert_flac()
+#copy_trs ('train-clean-360','trs_train-clean-360')
+#convert_speaker_ID_wav ('wav_train-clean-360')
+#link_wavs()
+#segments_speakers('wav_train-clean-360')
 text('trs_train-clean-360', 'wav_train-clean-360')
 lexicon('trs_train-clean-360')
-copy_phones()
+#copy_phones()
 
 
 

diff --git a/corpora/LibriSpeech_train-clean-360/bin/phones.txt b/corpora/LibriSpeech_train-clean-360/bin/phones.txt
@@ -0,0 +1,39 @@
+S s
+UW u
+T t
+N n
+K k
+Y j
+Z z
+AO ɔ
+AY aɪ
+SH ʃ
+W w
+NG ŋ
+EY e
+B b
+CH ʧ
+OY ɔɪ
+JH ʤ
+D d
+ZH ʒ
+G g
+UH ʊ
+F f
+V v
+ER ɝ
+AA ɑ
+IH ɪ
+M m
+DH ð
+L l
+AH ʌ
+P p
+OW o
+AW aʊ
+HH h
+AE æ
+R r
+TH θ
+IY i
+EH ɛ