github-actions[bot] commited on
Commit
53c941c
·
1 Parent(s): 006f88d

Update evaluation results [skip ci]

Browse files
Files changed (1) hide show
  1. languages.json +202 -202
languages.json CHANGED
@@ -269,11 +269,11 @@
269
  "language_name":"Wu Chinese",
270
  "autonym":"Wu Chinese",
271
  "family":"Sino-Tibetan",
272
- "flores_path":null,
273
  "fleurs_tag":null,
274
  "commonvoice_hours":null,
275
  "commonvoice_locale":null,
276
- "in_benchmark":false
277
  },
278
  {
279
  "bcp_47":"tr",
@@ -485,11 +485,11 @@
485
  "language_name":"North Levantine Arabic",
486
  "autonym":"العامية",
487
  "family":"Afro-Asiatic",
488
- "flores_path":null,
489
  "fleurs_tag":null,
490
  "commonvoice_hours":null,
491
  "commonvoice_locale":null,
492
- "in_benchmark":false
493
  },
494
  {
495
  "bcp_47":"ms",
@@ -876,10 +876,10 @@
876
  "in_benchmark":true
877
  },
878
  {
879
- "bcp_47":"bgc",
880
  "speakers":15913080,
881
- "language_name":"Haryanvi",
882
- "autonym":"हरियाणवी",
883
  "family":"Indo-European",
884
  "flores_path":null,
885
  "fleurs_tag":null,
@@ -888,10 +888,10 @@
888
  "in_benchmark":false
889
  },
890
  {
891
- "bcp_47":"mwr",
892
  "speakers":15913080,
893
- "language_name":"Marwari",
894
- "autonym":"Marwari",
895
  "family":"Indo-European",
896
  "flores_path":null,
897
  "fleurs_tag":null,
@@ -1073,11 +1073,11 @@
1073
  "language_name":"Akan",
1074
  "autonym":"Akan",
1075
  "family":"Atlantic-Congo",
1076
- "flores_path":null,
1077
  "fleurs_tag":null,
1078
  "commonvoice_hours":0.2,
1079
  "commonvoice_locale":"tw",
1080
- "in_benchmark":false
1081
  },
1082
  {
1083
  "bcp_47":"qu",
@@ -1955,18 +1955,6 @@
1955
  "commonvoice_locale":"gom",
1956
  "in_benchmark":true
1957
  },
1958
- {
1959
- "bcp_47":"kam",
1960
- "speakers":4068120,
1961
- "language_name":"Kamba",
1962
- "autonym":"Kikamba",
1963
- "family":"Atlantic-Congo",
1964
- "flores_path":"kam_Latn",
1965
- "fleurs_tag":"kam_ke",
1966
- "commonvoice_hours":0.0,
1967
- "commonvoice_locale":"kam",
1968
- "in_benchmark":true
1969
- },
1970
  {
1971
  "bcp_47":"kln",
1972
  "speakers":4068120,
@@ -1979,6 +1967,18 @@
1979
  "commonvoice_locale":"kln",
1980
  "in_benchmark":false
1981
  },
 
 
 
 
 
 
 
 
 
 
 
 
1982
  {
1983
  "bcp_47":"bjn",
1984
  "speakers":4010288,
@@ -2009,11 +2009,11 @@
2009
  "language_name":"Makhuwa",
2010
  "autonym":"Emakhuwa",
2011
  "family":"Atlantic-Congo",
2012
- "flores_path":null,
2013
  "fleurs_tag":null,
2014
  "commonvoice_hours":0.0,
2015
  "commonvoice_locale":"vmw",
2016
- "in_benchmark":false
2017
  },
2018
  {
2019
  "bcp_47":"glk",
@@ -2124,10 +2124,10 @@
2124
  "in_benchmark":true
2125
  },
2126
  {
2127
- "bcp_47":"lmn",
2128
  "speakers":3580443,
2129
- "language_name":"Lambadi",
2130
- "autonym":"Lambadi",
2131
  "family":"Indo-European",
2132
  "flores_path":null,
2133
  "fleurs_tag":null,
@@ -2136,10 +2136,10 @@
2136
  "in_benchmark":false
2137
  },
2138
  {
2139
- "bcp_47":"gbm",
2140
  "speakers":3580443,
2141
- "language_name":"Garhwali",
2142
- "autonym":"Garhwali",
2143
  "family":"Indo-European",
2144
  "flores_path":null,
2145
  "fleurs_tag":null,
@@ -2352,27 +2352,27 @@
2352
  "in_benchmark":true
2353
  },
2354
  {
2355
- "bcp_47":"ibb",
2356
  "speakers":2996392,
2357
- "language_name":"Ibibio",
2358
- "autonym":"Ibibio",
2359
  "family":"Atlantic-Congo",
2360
  "flores_path":null,
2361
  "fleurs_tag":null,
2362
- "commonvoice_hours":4.3,
2363
- "commonvoice_locale":"ibb",
2364
  "in_benchmark":false
2365
  },
2366
  {
2367
- "bcp_47":"efi",
2368
  "speakers":2996392,
2369
- "language_name":"Efik",
2370
- "autonym":"Efik",
2371
  "family":"Atlantic-Congo",
2372
  "flores_path":null,
2373
  "fleurs_tag":null,
2374
- "commonvoice_hours":null,
2375
- "commonvoice_locale":null,
2376
  "in_benchmark":false
2377
  },
2378
  {
@@ -2544,11 +2544,11 @@
2544
  "in_benchmark":false
2545
  },
2546
  {
2547
- "bcp_47":"sck",
2548
  "speakers":2386962,
2549
- "language_name":"Sadri",
2550
- "autonym":"Sadri",
2551
- "family":"Indo-European",
2552
  "flores_path":null,
2553
  "fleurs_tag":null,
2554
  "commonvoice_hours":null,
@@ -2556,11 +2556,11 @@
2556
  "in_benchmark":false
2557
  },
2558
  {
2559
- "bcp_47":"wbq",
2560
  "speakers":2386962,
2561
- "language_name":"Waddar",
2562
- "autonym":"Waddar",
2563
- "family":"Dravidian",
2564
  "flores_path":null,
2565
  "fleurs_tag":null,
2566
  "commonvoice_hours":null,
@@ -2724,10 +2724,10 @@
2724
  "in_benchmark":false
2725
  },
2726
  {
2727
- "bcp_47":"wbr",
2728
  "speakers":1989135,
2729
- "language_name":"Wagdi",
2730
- "autonym":"Wagdi",
2731
  "family":"Indo-European",
2732
  "flores_path":null,
2733
  "fleurs_tag":null,
@@ -2748,10 +2748,10 @@
2748
  "in_benchmark":false
2749
  },
2750
  {
2751
- "bcp_47":"khn",
2752
  "speakers":1989135,
2753
- "language_name":"Khandesi",
2754
- "autonym":"Khandesi",
2755
  "family":"Indo-European",
2756
  "flores_path":null,
2757
  "fleurs_tag":null,
@@ -3684,10 +3684,10 @@
3684
  "in_benchmark":true
3685
  },
3686
  {
3687
- "bcp_47":"bez",
3688
  "speakers":995398,
3689
- "language_name":"Bena",
3690
- "autonym":"Hibena",
3691
  "family":"Atlantic-Congo",
3692
  "flores_path":null,
3693
  "fleurs_tag":null,
@@ -3696,10 +3696,10 @@
3696
  "in_benchmark":false
3697
  },
3698
  {
3699
- "bcp_47":"ksb",
3700
  "speakers":995398,
3701
- "language_name":"Shambala",
3702
- "autonym":"Kishambaa",
3703
  "family":"Atlantic-Congo",
3704
  "flores_path":null,
3705
  "fleurs_tag":null,
@@ -4512,27 +4512,27 @@
4512
  "in_benchmark":false
4513
  },
4514
  {
4515
- "bcp_47":"nhe",
4516
  "speakers":501735,
4517
- "language_name":"Eastern Huasteca Nahuatl",
4518
- "autonym":"Eastern Huasteca Nahuatl",
4519
  "family":"Uto-Aztecan",
4520
  "flores_path":null,
4521
  "fleurs_tag":null,
4522
- "commonvoice_hours":0.0,
4523
- "commonvoice_locale":"nhe",
4524
  "in_benchmark":false
4525
  },
4526
  {
4527
- "bcp_47":"nhw",
4528
  "speakers":501735,
4529
- "language_name":"Western Huasteca Nahuatl",
4530
- "autonym":"Western Huasteca Nahuatl",
4531
  "family":"Uto-Aztecan",
4532
  "flores_path":null,
4533
  "fleurs_tag":null,
4534
- "commonvoice_hours":null,
4535
- "commonvoice_locale":null,
4536
  "in_benchmark":false
4537
  },
4538
  {
@@ -4685,11 +4685,11 @@
4685
  "language_name":"Erzya",
4686
  "autonym":"Эрзянь Кель",
4687
  "family":"Uralic",
4688
- "flores_path":null,
4689
  "fleurs_tag":null,
4690
  "commonvoice_hours":3.8,
4691
  "commonvoice_locale":"myv",
4692
- "in_benchmark":false
4693
  },
4694
  {
4695
  "bcp_47":"dav",
@@ -4715,6 +4715,18 @@
4715
  "commonvoice_locale":null,
4716
  "in_benchmark":false
4717
  },
 
 
 
 
 
 
 
 
 
 
 
 
4718
  {
4719
  "bcp_47":"vun",
4720
  "speakers":433291,
@@ -4739,18 +4751,6 @@
4739
  "commonvoice_locale":"rof",
4740
  "in_benchmark":false
4741
  },
4742
- {
4743
- "bcp_47":"jmc",
4744
- "speakers":433291,
4745
- "language_name":"Machame",
4746
- "autonym":"Kimachame",
4747
- "family":"Atlantic-Congo",
4748
- "flores_path":null,
4749
- "fleurs_tag":null,
4750
- "commonvoice_hours":null,
4751
- "commonvoice_locale":null,
4752
- "in_benchmark":false
4753
- },
4754
  {
4755
  "bcp_47":"kjg",
4756
  "speakers":431949,
@@ -5009,11 +5009,11 @@
5009
  "language_name":"Dargwa",
5010
  "autonym":"Dargwa",
5011
  "family":"Nakh-Daghestanian",
5012
- "flores_path":null,
5013
  "fleurs_tag":null,
5014
  "commonvoice_hours":0.0,
5015
  "commonvoice_locale":"dar",
5016
- "in_benchmark":false
5017
  },
5018
  {
5019
  "bcp_47":"lif",
@@ -5124,27 +5124,27 @@
5124
  "in_benchmark":false
5125
  },
5126
  {
5127
- "bcp_47":"bax",
5128
  "speakers":332940,
5129
- "language_name":"Bamun",
5130
- "autonym":"Bamun",
5131
  "family":"Atlantic-Congo",
5132
  "flores_path":null,
5133
  "fleurs_tag":null,
5134
- "commonvoice_hours":11.0,
5135
- "commonvoice_locale":"bax",
5136
  "in_benchmark":false
5137
  },
5138
  {
5139
- "bcp_47":"bas",
5140
  "speakers":332940,
5141
- "language_name":"Basaa",
5142
- "autonym":"Ɓàsàa",
5143
  "family":"Atlantic-Congo",
5144
  "flores_path":null,
5145
  "fleurs_tag":null,
5146
- "commonvoice_hours":12.0,
5147
- "commonvoice_locale":"bas",
5148
  "in_benchmark":false
5149
  },
5150
  {
@@ -5232,11 +5232,11 @@
5232
  "in_benchmark":false
5233
  },
5234
  {
5235
- "bcp_47":"njo",
5236
  "speakers":305001,
5237
- "language_name":"Ao Naga",
5238
- "autonym":"Ao Naga",
5239
- "family":"Sino-Tibetan",
5240
  "flores_path":null,
5241
  "fleurs_tag":null,
5242
  "commonvoice_hours":null,
@@ -5244,11 +5244,11 @@
5244
  "in_benchmark":false
5245
  },
5246
  {
5247
- "bcp_47":"bfq",
5248
  "speakers":305001,
5249
- "language_name":"Badaga",
5250
- "autonym":"Badaga",
5251
- "family":"Dravidian",
5252
  "flores_path":null,
5253
  "fleurs_tag":null,
5254
  "commonvoice_hours":null,
@@ -5388,10 +5388,10 @@
5388
  "in_benchmark":false
5389
  },
5390
  {
5391
- "bcp_47":"khb",
5392
  "speakers":264864,
5393
- "language_name":"",
5394
- "autonym":"",
5395
  "family":"Tai-Kadai",
5396
  "flores_path":null,
5397
  "fleurs_tag":null,
@@ -5400,10 +5400,10 @@
5400
  "in_benchmark":false
5401
  },
5402
  {
5403
- "bcp_47":"tdd",
5404
  "speakers":264864,
5405
- "language_name":"Tai Nüa",
5406
- "autonym":"Tai Nüa",
5407
  "family":"Tai-Kadai",
5408
  "flores_path":null,
5409
  "fleurs_tag":null,
@@ -5508,10 +5508,10 @@
5508
  "in_benchmark":true
5509
  },
5510
  {
5511
- "bcp_47":"mdr",
5512
  "speakers":245664,
5513
- "language_name":"Mandar",
5514
- "autonym":"Mandar",
5515
  "family":"Austronesian",
5516
  "flores_path":null,
5517
  "fleurs_tag":null,
@@ -5520,10 +5520,10 @@
5520
  "in_benchmark":false
5521
  },
5522
  {
5523
- "bcp_47":"sxn",
5524
  "speakers":245664,
5525
- "language_name":"Sangir",
5526
- "autonym":"Sangir",
5527
  "family":"Austronesian",
5528
  "flores_path":null,
5529
  "fleurs_tag":null,
@@ -5753,11 +5753,11 @@
5753
  "language_name":"Tuvinian",
5754
  "autonym":"Tuvinian",
5755
  "family":"Turkic",
5756
- "flores_path":null,
5757
  "fleurs_tag":null,
5758
  "commonvoice_hours":0.0,
5759
  "commonvoice_locale":"tyv",
5760
- "in_benchmark":false
5761
  },
5762
  {
5763
  "bcp_47":"dtp",
@@ -5904,10 +5904,10 @@
5904
  "in_benchmark":false
5905
  },
5906
  {
5907
- "bcp_47":"kkj",
5908
  "speakers":149823,
5909
- "language_name":"Kako",
5910
- "autonym":"Kakɔ",
5911
  "family":"Atlantic-Congo",
5912
  "flores_path":null,
5913
  "fleurs_tag":null,
@@ -5916,10 +5916,10 @@
5916
  "in_benchmark":false
5917
  },
5918
  {
5919
- "bcp_47":"bss",
5920
  "speakers":149823,
5921
- "language_name":"Akoose",
5922
- "autonym":"Akoose",
5923
  "family":"Atlantic-Congo",
5924
  "flores_path":null,
5925
  "fleurs_tag":null,
@@ -6821,11 +6821,11 @@
6821
  "language_name":"Aragonese",
6822
  "autonym":"Aragonés",
6823
  "family":"Indo-European",
6824
- "flores_path":null,
6825
  "fleurs_tag":null,
6826
  "commonvoice_hours":17.0,
6827
  "commonvoice_locale":"an",
6828
- "in_benchmark":false
6829
  },
6830
  {
6831
  "bcp_47":"chr",
@@ -7272,11 +7272,11 @@
7272
  "in_benchmark":false
7273
  },
7274
  {
7275
- "bcp_47":"twq",
7276
  "speakers":7970,
7277
- "language_name":"Tasawaq",
7278
- "autonym":"Tasawaq Senni",
7279
- "family":"Songhay",
7280
  "flores_path":null,
7281
  "fleurs_tag":null,
7282
  "commonvoice_hours":null,
@@ -7284,11 +7284,11 @@
7284
  "in_benchmark":false
7285
  },
7286
  {
7287
- "bcp_47":"bku",
7288
  "speakers":7970,
7289
- "language_name":"Buhid",
7290
- "autonym":"Buhid",
7291
- "family":"Austronesian",
7292
  "flores_path":null,
7293
  "fleurs_tag":null,
7294
  "commonvoice_hours":null,
@@ -7836,11 +7836,11 @@
7836
  "in_benchmark":false
7837
  },
7838
  {
7839
- "bcp_47":"kwk",
7840
  "speakers":377,
7841
- "language_name":"Kwakʼwala",
7842
- "autonym":"KwakʼWala",
7843
- "family":"Wakashan",
7844
  "flores_path":null,
7845
  "fleurs_tag":null,
7846
  "commonvoice_hours":null,
@@ -7848,11 +7848,11 @@
7848
  "in_benchmark":false
7849
  },
7850
  {
7851
- "bcp_47":"crl",
7852
  "speakers":377,
7853
- "language_name":"Northern East Cree",
7854
- "autonym":"Northern East Cree",
7855
- "family":"Algic",
7856
  "flores_path":null,
7857
  "fleurs_tag":null,
7858
  "commonvoice_hours":null,
@@ -7968,11 +7968,11 @@
7968
  "in_benchmark":false
7969
  },
7970
  {
7971
- "bcp_47":"lzh",
7972
  "speakers":0,
7973
- "language_name":"Literary Chinese",
7974
- "autonym":"Literary Chinese",
7975
- "family":"Sino-Tibetan",
7976
  "flores_path":null,
7977
  "fleurs_tag":null,
7978
  "commonvoice_hours":null,
@@ -7980,11 +7980,11 @@
7980
  "in_benchmark":false
7981
  },
7982
  {
7983
- "bcp_47":"io",
7984
  "speakers":0,
7985
- "language_name":"Ido",
7986
- "autonym":"Ido",
7987
- "family":"Artificial Language",
7988
  "flores_path":null,
7989
  "fleurs_tag":null,
7990
  "commonvoice_hours":null,
@@ -7992,22 +7992,22 @@
7992
  "in_benchmark":false
7993
  },
7994
  {
7995
- "bcp_47":"jbo",
7996
  "speakers":0,
7997
- "language_name":"Lojban",
7998
- "autonym":"La .Lojban.",
7999
- "family":"Artificial Language",
8000
  "flores_path":null,
8001
  "fleurs_tag":null,
8002
- "commonvoice_hours":0.0,
8003
- "commonvoice_locale":"jbo",
8004
  "in_benchmark":false
8005
  },
8006
  {
8007
- "bcp_47":"jut",
8008
  "speakers":0,
8009
- "language_name":"Jutish",
8010
- "autonym":"Jutish",
8011
  "family":"Indo-European",
8012
  "flores_path":null,
8013
  "fleurs_tag":null,
@@ -8016,23 +8016,23 @@
8016
  "in_benchmark":false
8017
  },
8018
  {
8019
- "bcp_47":"vot",
8020
  "speakers":0,
8021
- "language_name":"Votic",
8022
- "autonym":"Votic",
8023
- "family":"Uralic",
8024
  "flores_path":null,
8025
  "fleurs_tag":null,
8026
- "commonvoice_hours":0.1,
8027
- "commonvoice_locale":"vot",
8028
  "in_benchmark":false
8029
  },
8030
  {
8031
- "bcp_47":"gez",
8032
  "speakers":0,
8033
- "language_name":"Geez",
8034
- "autonym":"Geez",
8035
- "family":"Afro-Asiatic",
8036
  "flores_path":null,
8037
  "fleurs_tag":null,
8038
  "commonvoice_hours":null,
@@ -8040,23 +8040,23 @@
8040
  "in_benchmark":false
8041
  },
8042
  {
8043
- "bcp_47":"osa",
8044
  "speakers":0,
8045
- "language_name":"Osage",
8046
- "autonym":"𐓏𐓘𐓻𐓘𐓻𐓟",
8047
- "family":"Siouan",
8048
  "flores_path":null,
8049
  "fleurs_tag":null,
8050
- "commonvoice_hours":null,
8051
- "commonvoice_locale":null,
8052
  "in_benchmark":false
8053
  },
8054
  {
8055
- "bcp_47":"rgn",
8056
  "speakers":0,
8057
- "language_name":"Romagnol",
8058
- "autonym":"Romagnol",
8059
- "family":"Indo-European",
8060
  "flores_path":null,
8061
  "fleurs_tag":null,
8062
  "commonvoice_hours":null,
@@ -8064,10 +8064,10 @@
8064
  "in_benchmark":false
8065
  },
8066
  {
8067
- "bcp_47":"cu",
8068
  "speakers":0,
8069
- "language_name":"Church Slavic",
8070
- "autonym":"Church Slavic",
8071
  "family":"Indo-European",
8072
  "flores_path":null,
8073
  "fleurs_tag":null,
@@ -8076,11 +8076,11 @@
8076
  "in_benchmark":false
8077
  },
8078
  {
8079
- "bcp_47":"sgs",
8080
  "speakers":0,
8081
- "language_name":"Samogitian",
8082
- "autonym":"Samogitian",
8083
- "family":"Indo-European",
8084
  "flores_path":null,
8085
  "fleurs_tag":null,
8086
  "commonvoice_hours":null,
@@ -8088,11 +8088,11 @@
8088
  "in_benchmark":false
8089
  },
8090
  {
8091
- "bcp_47":"ann",
8092
  "speakers":0,
8093
- "language_name":"Obolo",
8094
- "autonym":"Obolo",
8095
- "family":"Atlantic-Congo",
8096
  "flores_path":null,
8097
  "fleurs_tag":null,
8098
  "commonvoice_hours":null,
@@ -8112,23 +8112,23 @@
8112
  "in_benchmark":false
8113
  },
8114
  {
8115
- "bcp_47":"cad",
8116
  "speakers":0,
8117
- "language_name":"Caddo",
8118
- "autonym":"Caddo",
8119
- "family":"Caddoan",
8120
  "flores_path":null,
8121
  "fleurs_tag":null,
8122
- "commonvoice_hours":null,
8123
- "commonvoice_locale":null,
8124
  "in_benchmark":false
8125
  },
8126
  {
8127
- "bcp_47":"pfl",
8128
  "speakers":0,
8129
- "language_name":"Palatine German",
8130
- "autonym":"Palatine German",
8131
- "family":"Indo-European",
8132
  "flores_path":null,
8133
  "fleurs_tag":null,
8134
  "commonvoice_hours":null,
 
269
  "language_name":"Wu Chinese",
270
  "autonym":"Wu Chinese",
271
  "family":"Sino-Tibetan",
272
+ "flores_path":"wuu_Hans",
273
  "fleurs_tag":null,
274
  "commonvoice_hours":null,
275
  "commonvoice_locale":null,
276
+ "in_benchmark":true
277
  },
278
  {
279
  "bcp_47":"tr",
 
485
  "language_name":"North Levantine Arabic",
486
  "autonym":"العامية",
487
  "family":"Afro-Asiatic",
488
+ "flores_path":"apc_Arab",
489
  "fleurs_tag":null,
490
  "commonvoice_hours":null,
491
  "commonvoice_locale":null,
492
+ "in_benchmark":true
493
  },
494
  {
495
  "bcp_47":"ms",
 
876
  "in_benchmark":true
877
  },
878
  {
879
+ "bcp_47":"mwr",
880
  "speakers":15913080,
881
+ "language_name":"Marwari",
882
+ "autonym":"Marwari",
883
  "family":"Indo-European",
884
  "flores_path":null,
885
  "fleurs_tag":null,
 
888
  "in_benchmark":false
889
  },
890
  {
891
+ "bcp_47":"bgc",
892
  "speakers":15913080,
893
+ "language_name":"Haryanvi",
894
+ "autonym":"हरियाणवी",
895
  "family":"Indo-European",
896
  "flores_path":null,
897
  "fleurs_tag":null,
 
1073
  "language_name":"Akan",
1074
  "autonym":"Akan",
1075
  "family":"Atlantic-Congo",
1076
+ "flores_path":"twi_Latn",
1077
  "fleurs_tag":null,
1078
  "commonvoice_hours":0.2,
1079
  "commonvoice_locale":"tw",
1080
+ "in_benchmark":true
1081
  },
1082
  {
1083
  "bcp_47":"qu",
 
1955
  "commonvoice_locale":"gom",
1956
  "in_benchmark":true
1957
  },
 
 
 
 
 
 
 
 
 
 
 
 
1958
  {
1959
  "bcp_47":"kln",
1960
  "speakers":4068120,
 
1967
  "commonvoice_locale":"kln",
1968
  "in_benchmark":false
1969
  },
1970
+ {
1971
+ "bcp_47":"kam",
1972
+ "speakers":4068120,
1973
+ "language_name":"Kamba",
1974
+ "autonym":"Kikamba",
1975
+ "family":"Atlantic-Congo",
1976
+ "flores_path":"kam_Latn",
1977
+ "fleurs_tag":"kam_ke",
1978
+ "commonvoice_hours":0.0,
1979
+ "commonvoice_locale":"kam",
1980
+ "in_benchmark":true
1981
+ },
1982
  {
1983
  "bcp_47":"bjn",
1984
  "speakers":4010288,
 
2009
  "language_name":"Makhuwa",
2010
  "autonym":"Emakhuwa",
2011
  "family":"Atlantic-Congo",
2012
+ "flores_path":"vmw_Latn",
2013
  "fleurs_tag":null,
2014
  "commonvoice_hours":0.0,
2015
  "commonvoice_locale":"vmw",
2016
+ "in_benchmark":true
2017
  },
2018
  {
2019
  "bcp_47":"glk",
 
2124
  "in_benchmark":true
2125
  },
2126
  {
2127
+ "bcp_47":"gbm",
2128
  "speakers":3580443,
2129
+ "language_name":"Garhwali",
2130
+ "autonym":"Garhwali",
2131
  "family":"Indo-European",
2132
  "flores_path":null,
2133
  "fleurs_tag":null,
 
2136
  "in_benchmark":false
2137
  },
2138
  {
2139
+ "bcp_47":"lmn",
2140
  "speakers":3580443,
2141
+ "language_name":"Lambadi",
2142
+ "autonym":"Lambadi",
2143
  "family":"Indo-European",
2144
  "flores_path":null,
2145
  "fleurs_tag":null,
 
2352
  "in_benchmark":true
2353
  },
2354
  {
2355
+ "bcp_47":"efi",
2356
  "speakers":2996392,
2357
+ "language_name":"Efik",
2358
+ "autonym":"Efik",
2359
  "family":"Atlantic-Congo",
2360
  "flores_path":null,
2361
  "fleurs_tag":null,
2362
+ "commonvoice_hours":null,
2363
+ "commonvoice_locale":null,
2364
  "in_benchmark":false
2365
  },
2366
  {
2367
+ "bcp_47":"ibb",
2368
  "speakers":2996392,
2369
+ "language_name":"Ibibio",
2370
+ "autonym":"Ibibio",
2371
  "family":"Atlantic-Congo",
2372
  "flores_path":null,
2373
  "fleurs_tag":null,
2374
+ "commonvoice_hours":4.3,
2375
+ "commonvoice_locale":"ibb",
2376
  "in_benchmark":false
2377
  },
2378
  {
 
2544
  "in_benchmark":false
2545
  },
2546
  {
2547
+ "bcp_47":"wbq",
2548
  "speakers":2386962,
2549
+ "language_name":"Waddar",
2550
+ "autonym":"Waddar",
2551
+ "family":"Dravidian",
2552
  "flores_path":null,
2553
  "fleurs_tag":null,
2554
  "commonvoice_hours":null,
 
2556
  "in_benchmark":false
2557
  },
2558
  {
2559
+ "bcp_47":"sck",
2560
  "speakers":2386962,
2561
+ "language_name":"Sadri",
2562
+ "autonym":"Sadri",
2563
+ "family":"Indo-European",
2564
  "flores_path":null,
2565
  "fleurs_tag":null,
2566
  "commonvoice_hours":null,
 
2724
  "in_benchmark":false
2725
  },
2726
  {
2727
+ "bcp_47":"khn",
2728
  "speakers":1989135,
2729
+ "language_name":"Khandesi",
2730
+ "autonym":"Khandesi",
2731
  "family":"Indo-European",
2732
  "flores_path":null,
2733
  "fleurs_tag":null,
 
2748
  "in_benchmark":false
2749
  },
2750
  {
2751
+ "bcp_47":"wbr",
2752
  "speakers":1989135,
2753
+ "language_name":"Wagdi",
2754
+ "autonym":"Wagdi",
2755
  "family":"Indo-European",
2756
  "flores_path":null,
2757
  "fleurs_tag":null,
 
3684
  "in_benchmark":true
3685
  },
3686
  {
3687
+ "bcp_47":"ksb",
3688
  "speakers":995398,
3689
+ "language_name":"Shambala",
3690
+ "autonym":"Kishambaa",
3691
  "family":"Atlantic-Congo",
3692
  "flores_path":null,
3693
  "fleurs_tag":null,
 
3696
  "in_benchmark":false
3697
  },
3698
  {
3699
+ "bcp_47":"bez",
3700
  "speakers":995398,
3701
+ "language_name":"Bena",
3702
+ "autonym":"Hibena",
3703
  "family":"Atlantic-Congo",
3704
  "flores_path":null,
3705
  "fleurs_tag":null,
 
4512
  "in_benchmark":false
4513
  },
4514
  {
4515
+ "bcp_47":"nhw",
4516
  "speakers":501735,
4517
+ "language_name":"Western Huasteca Nahuatl",
4518
+ "autonym":"Western Huasteca Nahuatl",
4519
  "family":"Uto-Aztecan",
4520
  "flores_path":null,
4521
  "fleurs_tag":null,
4522
+ "commonvoice_hours":null,
4523
+ "commonvoice_locale":null,
4524
  "in_benchmark":false
4525
  },
4526
  {
4527
+ "bcp_47":"nhe",
4528
  "speakers":501735,
4529
+ "language_name":"Eastern Huasteca Nahuatl",
4530
+ "autonym":"Eastern Huasteca Nahuatl",
4531
  "family":"Uto-Aztecan",
4532
  "flores_path":null,
4533
  "fleurs_tag":null,
4534
+ "commonvoice_hours":0.0,
4535
+ "commonvoice_locale":"nhe",
4536
  "in_benchmark":false
4537
  },
4538
  {
 
4685
  "language_name":"Erzya",
4686
  "autonym":"Эрзянь Кель",
4687
  "family":"Uralic",
4688
+ "flores_path":"myv_Cyrl",
4689
  "fleurs_tag":null,
4690
  "commonvoice_hours":3.8,
4691
  "commonvoice_locale":"myv",
4692
+ "in_benchmark":true
4693
  },
4694
  {
4695
  "bcp_47":"dav",
 
4715
  "commonvoice_locale":null,
4716
  "in_benchmark":false
4717
  },
4718
+ {
4719
+ "bcp_47":"jmc",
4720
+ "speakers":433291,
4721
+ "language_name":"Machame",
4722
+ "autonym":"Kimachame",
4723
+ "family":"Atlantic-Congo",
4724
+ "flores_path":null,
4725
+ "fleurs_tag":null,
4726
+ "commonvoice_hours":null,
4727
+ "commonvoice_locale":null,
4728
+ "in_benchmark":false
4729
+ },
4730
  {
4731
  "bcp_47":"vun",
4732
  "speakers":433291,
 
4751
  "commonvoice_locale":"rof",
4752
  "in_benchmark":false
4753
  },
 
 
 
 
 
 
 
 
 
 
 
 
4754
  {
4755
  "bcp_47":"kjg",
4756
  "speakers":431949,
 
5009
  "language_name":"Dargwa",
5010
  "autonym":"Dargwa",
5011
  "family":"Nakh-Daghestanian",
5012
+ "flores_path":"dar_Cyrl",
5013
  "fleurs_tag":null,
5014
  "commonvoice_hours":0.0,
5015
  "commonvoice_locale":"dar",
5016
+ "in_benchmark":true
5017
  },
5018
  {
5019
  "bcp_47":"lif",
 
5124
  "in_benchmark":false
5125
  },
5126
  {
5127
+ "bcp_47":"bas",
5128
  "speakers":332940,
5129
+ "language_name":"Basaa",
5130
+ "autonym":"Ɓàsàa",
5131
  "family":"Atlantic-Congo",
5132
  "flores_path":null,
5133
  "fleurs_tag":null,
5134
+ "commonvoice_hours":12.0,
5135
+ "commonvoice_locale":"bas",
5136
  "in_benchmark":false
5137
  },
5138
  {
5139
+ "bcp_47":"bax",
5140
  "speakers":332940,
5141
+ "language_name":"Bamun",
5142
+ "autonym":"Bamun",
5143
  "family":"Atlantic-Congo",
5144
  "flores_path":null,
5145
  "fleurs_tag":null,
5146
+ "commonvoice_hours":11.0,
5147
+ "commonvoice_locale":"bax",
5148
  "in_benchmark":false
5149
  },
5150
  {
 
5232
  "in_benchmark":false
5233
  },
5234
  {
5235
+ "bcp_47":"bfq",
5236
  "speakers":305001,
5237
+ "language_name":"Badaga",
5238
+ "autonym":"Badaga",
5239
+ "family":"Dravidian",
5240
  "flores_path":null,
5241
  "fleurs_tag":null,
5242
  "commonvoice_hours":null,
 
5244
  "in_benchmark":false
5245
  },
5246
  {
5247
+ "bcp_47":"njo",
5248
  "speakers":305001,
5249
+ "language_name":"Ao Naga",
5250
+ "autonym":"Ao Naga",
5251
+ "family":"Sino-Tibetan",
5252
  "flores_path":null,
5253
  "fleurs_tag":null,
5254
  "commonvoice_hours":null,
 
5388
  "in_benchmark":false
5389
  },
5390
  {
5391
+ "bcp_47":"tdd",
5392
  "speakers":264864,
5393
+ "language_name":"Tai Nüa",
5394
+ "autonym":"Tai Nüa",
5395
  "family":"Tai-Kadai",
5396
  "flores_path":null,
5397
  "fleurs_tag":null,
 
5400
  "in_benchmark":false
5401
  },
5402
  {
5403
+ "bcp_47":"khb",
5404
  "speakers":264864,
5405
+ "language_name":"",
5406
+ "autonym":"",
5407
  "family":"Tai-Kadai",
5408
  "flores_path":null,
5409
  "fleurs_tag":null,
 
5508
  "in_benchmark":true
5509
  },
5510
  {
5511
+ "bcp_47":"sxn",
5512
  "speakers":245664,
5513
+ "language_name":"Sangir",
5514
+ "autonym":"Sangir",
5515
  "family":"Austronesian",
5516
  "flores_path":null,
5517
  "fleurs_tag":null,
 
5520
  "in_benchmark":false
5521
  },
5522
  {
5523
+ "bcp_47":"mdr",
5524
  "speakers":245664,
5525
+ "language_name":"Mandar",
5526
+ "autonym":"Mandar",
5527
  "family":"Austronesian",
5528
  "flores_path":null,
5529
  "fleurs_tag":null,
 
5753
  "language_name":"Tuvinian",
5754
  "autonym":"Tuvinian",
5755
  "family":"Turkic",
5756
+ "flores_path":"tyv_Cyrl",
5757
  "fleurs_tag":null,
5758
  "commonvoice_hours":0.0,
5759
  "commonvoice_locale":"tyv",
5760
+ "in_benchmark":true
5761
  },
5762
  {
5763
  "bcp_47":"dtp",
 
5904
  "in_benchmark":false
5905
  },
5906
  {
5907
+ "bcp_47":"bss",
5908
  "speakers":149823,
5909
+ "language_name":"Akoose",
5910
+ "autonym":"Akoose",
5911
  "family":"Atlantic-Congo",
5912
  "flores_path":null,
5913
  "fleurs_tag":null,
 
5916
  "in_benchmark":false
5917
  },
5918
  {
5919
+ "bcp_47":"kkj",
5920
  "speakers":149823,
5921
+ "language_name":"Kako",
5922
+ "autonym":"Kakɔ",
5923
  "family":"Atlantic-Congo",
5924
  "flores_path":null,
5925
  "fleurs_tag":null,
 
6821
  "language_name":"Aragonese",
6822
  "autonym":"Aragonés",
6823
  "family":"Indo-European",
6824
+ "flores_path":"arg_Latn",
6825
  "fleurs_tag":null,
6826
  "commonvoice_hours":17.0,
6827
  "commonvoice_locale":"an",
6828
+ "in_benchmark":true
6829
  },
6830
  {
6831
  "bcp_47":"chr",
 
7272
  "in_benchmark":false
7273
  },
7274
  {
7275
+ "bcp_47":"bku",
7276
  "speakers":7970,
7277
+ "language_name":"Buhid",
7278
+ "autonym":"Buhid",
7279
+ "family":"Austronesian",
7280
  "flores_path":null,
7281
  "fleurs_tag":null,
7282
  "commonvoice_hours":null,
 
7284
  "in_benchmark":false
7285
  },
7286
  {
7287
+ "bcp_47":"twq",
7288
  "speakers":7970,
7289
+ "language_name":"Tasawaq",
7290
+ "autonym":"Tasawaq Senni",
7291
+ "family":"Songhay",
7292
  "flores_path":null,
7293
  "fleurs_tag":null,
7294
  "commonvoice_hours":null,
 
7836
  "in_benchmark":false
7837
  },
7838
  {
7839
+ "bcp_47":"crl",
7840
  "speakers":377,
7841
+ "language_name":"Northern East Cree",
7842
+ "autonym":"Northern East Cree",
7843
+ "family":"Algic",
7844
  "flores_path":null,
7845
  "fleurs_tag":null,
7846
  "commonvoice_hours":null,
 
7848
  "in_benchmark":false
7849
  },
7850
  {
7851
+ "bcp_47":"kwk",
7852
  "speakers":377,
7853
+ "language_name":"Kwakʼwala",
7854
+ "autonym":"KwakʼWala",
7855
+ "family":"Wakashan",
7856
  "flores_path":null,
7857
  "fleurs_tag":null,
7858
  "commonvoice_hours":null,
 
7968
  "in_benchmark":false
7969
  },
7970
  {
7971
+ "bcp_47":"sgs",
7972
  "speakers":0,
7973
+ "language_name":"Samogitian",
7974
+ "autonym":"Samogitian",
7975
+ "family":"Indo-European",
7976
  "flores_path":null,
7977
  "fleurs_tag":null,
7978
  "commonvoice_hours":null,
 
7980
  "in_benchmark":false
7981
  },
7982
  {
7983
+ "bcp_47":"rgn",
7984
  "speakers":0,
7985
+ "language_name":"Romagnol",
7986
+ "autonym":"Romagnol",
7987
+ "family":"Indo-European",
7988
  "flores_path":null,
7989
  "fleurs_tag":null,
7990
  "commonvoice_hours":null,
 
7992
  "in_benchmark":false
7993
  },
7994
  {
7995
+ "bcp_47":"ann",
7996
  "speakers":0,
7997
+ "language_name":"Obolo",
7998
+ "autonym":"Obolo",
7999
+ "family":"Atlantic-Congo",
8000
  "flores_path":null,
8001
  "fleurs_tag":null,
8002
+ "commonvoice_hours":null,
8003
+ "commonvoice_locale":null,
8004
  "in_benchmark":false
8005
  },
8006
  {
8007
+ "bcp_47":"pfl",
8008
  "speakers":0,
8009
+ "language_name":"Palatine German",
8010
+ "autonym":"Palatine German",
8011
  "family":"Indo-European",
8012
  "flores_path":null,
8013
  "fleurs_tag":null,
 
8016
  "in_benchmark":false
8017
  },
8018
  {
8019
+ "bcp_47":"osa",
8020
  "speakers":0,
8021
+ "language_name":"Osage",
8022
+ "autonym":"𐓏𐓘𐓻𐓘𐓻𐓟",
8023
+ "family":"Siouan",
8024
  "flores_path":null,
8025
  "fleurs_tag":null,
8026
+ "commonvoice_hours":null,
8027
+ "commonvoice_locale":null,
8028
  "in_benchmark":false
8029
  },
8030
  {
8031
+ "bcp_47":"lzh",
8032
  "speakers":0,
8033
+ "language_name":"Literary Chinese",
8034
+ "autonym":"Literary Chinese",
8035
+ "family":"Sino-Tibetan",
8036
  "flores_path":null,
8037
  "fleurs_tag":null,
8038
  "commonvoice_hours":null,
 
8040
  "in_benchmark":false
8041
  },
8042
  {
8043
+ "bcp_47":"jbo",
8044
  "speakers":0,
8045
+ "language_name":"Lojban",
8046
+ "autonym":"La .Lojban.",
8047
+ "family":"Artificial Language",
8048
  "flores_path":null,
8049
  "fleurs_tag":null,
8050
+ "commonvoice_hours":0.0,
8051
+ "commonvoice_locale":"jbo",
8052
  "in_benchmark":false
8053
  },
8054
  {
8055
+ "bcp_47":"io",
8056
  "speakers":0,
8057
+ "language_name":"Ido",
8058
+ "autonym":"Ido",
8059
+ "family":"Artificial Language",
8060
  "flores_path":null,
8061
  "fleurs_tag":null,
8062
  "commonvoice_hours":null,
 
8064
  "in_benchmark":false
8065
  },
8066
  {
8067
+ "bcp_47":"jut",
8068
  "speakers":0,
8069
+ "language_name":"Jutish",
8070
+ "autonym":"Jutish",
8071
  "family":"Indo-European",
8072
  "flores_path":null,
8073
  "fleurs_tag":null,
 
8076
  "in_benchmark":false
8077
  },
8078
  {
8079
+ "bcp_47":"gez",
8080
  "speakers":0,
8081
+ "language_name":"Geez",
8082
+ "autonym":"Geez",
8083
+ "family":"Afro-Asiatic",
8084
  "flores_path":null,
8085
  "fleurs_tag":null,
8086
  "commonvoice_hours":null,
 
8088
  "in_benchmark":false
8089
  },
8090
  {
8091
+ "bcp_47":"cu",
8092
  "speakers":0,
8093
+ "language_name":"Church Slavic",
8094
+ "autonym":"Church Slavic",
8095
+ "family":"Indo-European",
8096
  "flores_path":null,
8097
  "fleurs_tag":null,
8098
  "commonvoice_hours":null,
 
8112
  "in_benchmark":false
8113
  },
8114
  {
8115
+ "bcp_47":"vot",
8116
  "speakers":0,
8117
+ "language_name":"Votic",
8118
+ "autonym":"Votic",
8119
+ "family":"Uralic",
8120
  "flores_path":null,
8121
  "fleurs_tag":null,
8122
+ "commonvoice_hours":0.1,
8123
+ "commonvoice_locale":"vot",
8124
  "in_benchmark":false
8125
  },
8126
  {
8127
+ "bcp_47":"cad",
8128
  "speakers":0,
8129
+ "language_name":"Caddo",
8130
+ "autonym":"Caddo",
8131
+ "family":"Caddoan",
8132
  "flores_path":null,
8133
  "fleurs_tag":null,
8134
  "commonvoice_hours":null,