Hugging Face's logo

Spaces:
oza75
/
bambara-asr
Sleeping

App Files Community
1
Aboubacar OUATTARA - kaira commited on
Commit
589123e
•
1 Parent(s): 54440ac

add examples files

Browse files
Files changed (10) hide show
  1. bambara_utils.py +4 -0
  2. examples/female_rfi_ma_001.wav +3 -0
  3. examples/female_rfi_ma_002.wav +3 -0
  4. examples/female_rfi_ma_003.wav +3 -0
  5. examples/female_rfi_ma_004.wav +3 -0
  6. examples/female_rfi_ma_005.wav +3 -0
  7. examples/male_rfi_ma_001.wav +3 -0
  8. examples/male_rfi_ma_002.wav +3 -0
  9. examples/male_rfi_ma_003.wav +3 -0
  10. examples/male_rfi_ma_004.wav +3 -0
bambara_utils.py CHANGED
@@ -2,10 +2,14 @@ from typing import List
2
 
3
  from tokenizers import AddedToken
4
  from transformers import WhisperTokenizer, WhisperProcessor
 
5
  from transformers.models.whisper.tokenization_whisper import TO_LANGUAGE_CODE, TASK_IDS
6
 
7
  CUSTOM_TO_LANGUAGE_CODE = {**TO_LANGUAGE_CODE, "bambara": "bm"}
8
 
 
 
 
9
 
10
  class BambaraWhisperTokenizer(WhisperTokenizer):
11
  def __init__(self, *args, **kwargs):
 
2
 
3
  from tokenizers import AddedToken
4
  from transformers import WhisperTokenizer, WhisperProcessor
5
+ import transformers.models.whisper.tokenization_whisper as whisper_tokenization
6
  from transformers.models.whisper.tokenization_whisper import TO_LANGUAGE_CODE, TASK_IDS
7
 
8
  CUSTOM_TO_LANGUAGE_CODE = {**TO_LANGUAGE_CODE, "bambara": "bm"}
9
 
10
+ # Note: We update the whisper tokenizer constants. Not ideal but at least it works
11
+ whisper_tokenization.TO_LANGUAGE_CODE.update(CUSTOM_TO_LANGUAGE_CODE)
12
+
13
 
14
  class BambaraWhisperTokenizer(WhisperTokenizer):
15
  def __init__(self, *args, **kwargs):
examples/female_rfi_ma_001.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a28d56e6e897ac26bd888feb123e4e165ed922e0abafaa289b639683c932a62e
3
+ size 236112
examples/female_rfi_ma_002.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d48b74bee5d5d1cf631db0df2f1fda89864fc8894aab5f4866f0a74bc59d7c11
3
+ size 217634
examples/female_rfi_ma_003.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe65971b4a6bd183e6f97a8645e4c95cd3fb51773182d3cd7610925bbedef029
3
+ size 368544
examples/female_rfi_ma_004.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4a3b64887baf586f851e029ddc9aec3161045dc0722c9cb62cfe90da205ec8f
3
+ size 194526
examples/female_rfi_ma_005.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10e3584477b4e2e2810989232b9beb9f7fbf5c1ceedab4daff6e03eac8dc3995
3
+ size 405016
examples/male_rfi_ma_001.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fcafbb32b518cf2ecda97d0b91e1faf28533b21a6deb56c02ec3f15a608243b4
3
+ size 94022
examples/male_rfi_ma_002.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:23ad88e15548e2b19189fea1d68e2dd23d6223983e56b82a942d089cc4e8e1f1
3
+ size 168594
examples/male_rfi_ma_003.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc2a4a2412cfd90001d121f88609e506d048b56ae9b41615145fe6183984eb7e
3
+ size 84540
examples/male_rfi_ma_004.wav ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ee14344bbf20b9ae32f8d32e03504578ce8f885d805de18fa1466ad16076d9c4
3
+ size 261734