Quillot Mathias / volia

Download as
Browse Code ยป

Commit 1d8b79f61492cbb28ce0f087f3a9dcea8546e06b

Authored by quillotm
1 parent 16251ebfaa
Exists in master

Changed to make simple the utt2sub and sub2utt commands. Was too long, no it cost less time.

Showing 1 changed file with 15 additions and 19 deletions Side-by-side Diff

... ... @@ -92,32 +92,30 @@
92 92 of.write(f"{kaldi_id} {masseffect_id}\n")
93 93  
94 94  
95   -def utt2sub(file: str, subfile: str, outfile: str):
  95 +def utt2sub(file: str, outfile: str):
96 96 data = core.data.read_id_values(file)
97 97 keys = [key for key in data]
98 98  
99   - data_sub = core.data.read_id_values(subfile)
100   - keys_sub = [key for key in data_sub]
101   -
102 99 with open(outfile, "w") as of:
103   - for key in keys:
104   - subkeys = [subkey for subkey in keys_sub if subkey.startswith(key)]
105   - subkeys_str = " ".join(subkeys)
  100 + key_2_subkeys = {}
  101 + for subkey in keys:
  102 + key = subkey.replace(" ", "").replace("\n", "").split("_")[:-1]
  103 + key_2_subkeys[key] = []
  104 + key_2_subkeys[key].append(subkey)
  105 +
  106 + for key in key_2_subkeys:
  107 + subkeys_str = " ".join(key_2_subkeys[key])
106 108 of.write(f"{key} {subkeys_str}\n")
107 109  
108 110  
109   -def sub2utt(file: str, subfile: str, outfile: str):
  111 +def sub2utt(file: str, outfile: str):
110 112 data = core.data.read_id_values(file)
111 113 keys = [key for key in data]
112 114  
113   - data_sub = core.data.read_id_values(subfile)
114   - keys_sub = [key for key in data_sub]
115   -
116 115 with open(outfile, "w") as of:
117   - for key in keys:
118   - subkeys = [subkey for subkey in keys_sub if subkey.startswith(key)]
119   - for subkey in subkeys:
120   - of.write(f"{subkey} {key}\n")
  116 + for subkey in keys:
  117 + key = subkey.replace(" ", "").replace("\n", "").split("_")[:-1]
  118 + of.write(f"{subkey} {key}\n")
121 119  
122 120  
123 121 if __name__ == '__main__':
124 122  
... ... @@ -162,15 +160,13 @@
162 160  
163 161 # Create utt2sub
164 162 parser_utt2sub = subparsers.add_parser("utt2sub", help="generate utt2sub file")
165   - parser_utt2sub.add_argument("--file", required=True, type=str, help="features, list or labels file with normal ids")
166   - parser_utt2sub.add_argument("--subfile", required=True, type=str, help="features, list or labels file with sub ids")
  163 + parser_utt2sub.add_argument("--file", required=True, type=str, help="features, list or labels file with sub ids")
167 164 parser_utt2sub.add_argument("--outfile", required=True, type=str, help="output file")
168 165 parser_utt2sub.set_defaults(which="utt2sub")
169 166  
170 167 # Create sub2utt
171 168 parser_sub2utt = subparsers.add_parser("sub2utt", help="generate sub2utt file")
172   - parser_sub2utt.add_argument("--file", required=True, type=str, help="features, list or labels file with normal ids")
173   - parser_sub2utt.add_argument("--subfile", required=True, type=str, help="features, list or labels file sub ids")
  169 + parser_sub2utt.add_argument("--file", required=True, type=str, help="features, list or labels file with sub ids")
174 170 parser_sub2utt.add_argument("--outfile", required=True, type=str, help="output file")
175 171 parser_sub2utt.set_defaults(which="sub2utt")
176 172