add utt2dur generation function to the data module

Quillot Mathias
1 parent 3b5a487de8
Showing 1 changed file with 31 additions and 1 deletions Side-by-side Diff
volia/data.py
@@ -8,7 +8,6 @@
  
  
  
-
 def filter_file(file, filter, outfile):
     file_path = file
     filter_path = filter
  
  
@@ -24,9 +23,29 @@
  
  
 def convert(file, type_from, type):
+
     pass
  
+def utt2dur(wavscp: str, outfile: str):
+    
+    import wave
+    import contextlib
  
+    with open(wavscp, "r") as f, open(outfile, "w") as of:
+        for line in f:
+            splited = line.replace("\n", "").split(" ")
+            id_ = splited[0]
+            wav_ = splited[1]
+            duration = 0
+            
+            with contextlib.closing(wave.open(wav_,'r')) as wav_f:
+                frames = wav_f.getnframes()
+                rate = wav_f.getframerate()
+                duration = frames / float(rate)
+            
+            of.write(f"{id_} {duration}\n")
+
+
 if __name__ == "__main__":
     # Main parser
     parser = argparse.ArgumentParser(description="manage files")
@@ -46,6 +65,16 @@
     parser_convert.add_argument("--type", choices=["old-masseffect", "new-masseffect"], required=True)
     parser_convert.set_defaults(which="convert")
  
+    # utt2dur
+    parser_utt2dur = subparsers.add_parser("utt2dur", help="generate utt2dur file")
+    parser_utt2dur.add_argument("--wavscp", type=str, help="wav file", required=True)
+    parser_utt2dur.add_argument("--outfile", type=str, default="utt2dur", help="output file")
+    parser_utt2dur.set_defaults(which="utt2dur")
+
+    # TODO: utt2label_to_label2utt
+
+    # TODO: label2utt_to_utt2label
+
     # Parse
     args = parser.parse_args()
  
@@ -53,6 +82,7 @@
     runner = SubCommandRunner({
         "convert" : convert,
         "filter": filter_file,
+        "utt2dur": utt2dur
     })
  
     runner.run(args.which, args.__dict__, remove="which")
...	...	@@ -8,7 +8,6 @@
8	8
9	9
10	10
11		-
12	11	def filter_file(file, filter, outfile):
13	12	file_path = file
14	13	filter_path = filter
15	14
16	15
...	...	@@ -24,9 +23,29 @@
24	23
25	24
26	25	def convert(file, type_from, type):
	26	+
27	27	pass
28	28
	29	+def utt2dur(wavscp: str, outfile: str):
	30	+
	31	+ import wave
	32	+ import contextlib
29	33
	34	+ with open(wavscp, "r") as f, open(outfile, "w") as of:
	35	+ for line in f:
	36	+ splited = line.replace("\n", "").split(" ")
	37	+ id_ = splited[0]
	38	+ wav_ = splited[1]
	39	+ duration = 0
	40	+
	41	+ with contextlib.closing(wave.open(wav_,'r')) as wav_f:
	42	+ frames = wav_f.getnframes()
	43	+ rate = wav_f.getframerate()
	44	+ duration = frames / float(rate)
	45	+
	46	+ of.write(f"{id_} {duration}\n")
	47	+
	48	+
30	49	if __name__ == "__main__":
31	50	# Main parser
32	51	parser = argparse.ArgumentParser(description="manage files")
...	...	@@ -46,6 +65,16 @@
46	65	parser_convert.add_argument("--type", choices=["old-masseffect", "new-masseffect"], required=True)
47	66	parser_convert.set_defaults(which="convert")
48	67
	68	+ # utt2dur
	69	+ parser_utt2dur = subparsers.add_parser("utt2dur", help="generate utt2dur file")
	70	+ parser_utt2dur.add_argument("--wavscp", type=str, help="wav file", required=True)
	71	+ parser_utt2dur.add_argument("--outfile", type=str, default="utt2dur", help="output file")
	72	+ parser_utt2dur.set_defaults(which="utt2dur")
	73	+
	74	+ # TODO: utt2label_to_label2utt
	75	+
	76	+ # TODO: label2utt_to_utt2label
	77	+
49	78	# Parse
50	79	args = parser.parse_args()
51	80
...	...	@@ -53,6 +82,7 @@
53	82	runner = SubCommandRunner({
54	83	"convert" : convert,
55	84	"filter": filter_file,
	85	+ "utt2dur": utt2dur
56	86	})
57	87
58	88	runner.run(args.which, args.__dict__, remove="which")