rohanmiriyala commited on
Commit
66a9507
·
verified ·
1 Parent(s): d82e27c
Files changed (1) hide show
  1. lang_list.py +64 -0
lang_list.py ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Language dict
2
+ language_code_to_name = {
3
+ "asm": "Assamese",
4
+ "ben": "Bengali",
5
+ "guj": "Gujarati",
6
+ "hin": "Hindi",
7
+ "kan": "Kannada",
8
+ "mal": "Malayalam",
9
+ "mar": "Marathi",
10
+ "ory": "Odia",
11
+ "pan": "Punjabi",
12
+ "tam": "Tamil",
13
+ "tel": "Telugu",
14
+ "urd": "Urdu",
15
+ "eng": "English"
16
+
17
+ }
18
+ LANGUAGE_NAME_TO_CODE = {v: k for k, v in language_code_to_name.items()}
19
+
20
+ # Source langs: S2ST / S2TT / ASR don't need source lang
21
+ # T2TT / T2ST use this
22
+ text_source_language_codes = [
23
+ "asm",
24
+ "ben",
25
+ "guj",
26
+ "hin",
27
+ "kan",
28
+ "mal",
29
+ "mar",
30
+ "ory",
31
+ "pan",
32
+ "tam",
33
+ "tel",
34
+ "urd",
35
+ "eng"
36
+
37
+ ]
38
+ TEXT_SOURCE_LANGUAGE_NAMES = sorted([language_code_to_name[code] for code in text_source_language_codes])
39
+
40
+ # Target langs:
41
+ # S2ST / T2ST
42
+ s2st_target_language_codes = [
43
+ "asm",
44
+ "ben",
45
+ "guj",
46
+ "hin",
47
+ "kan",
48
+ "mal",
49
+ "mar",
50
+ "ory",
51
+ "pan",
52
+ "tam",
53
+ "tel",
54
+ "urd",
55
+ "eng"
56
+
57
+ ]
58
+ S2ST_TARGET_LANGUAGE_NAMES = sorted([language_code_to_name[code] for code in s2st_target_language_codes])
59
+ T2ST_TARGET_LANGUAGE_NAMES = S2ST_TARGET_LANGUAGE_NAMES
60
+
61
+ # S2TT / T2TT / ASR
62
+ S2TT_TARGET_LANGUAGE_NAMES = TEXT_SOURCE_LANGUAGE_NAMES
63
+ T2TT_TARGET_LANGUAGE_NAMES = TEXT_SOURCE_LANGUAGE_NAMES
64
+ ASR_TARGET_LANGUAGE_NAMES = TEXT_SOURCE_LANGUAGE_NAMES