Upload KotobaWhisperPipeline
Browse files- README.md +44 -47
- config.json +1 -1
- generation_config.json +1 -1
- kotoba_whisper.py +1 -1
README.md
CHANGED
@@ -1,62 +1,59 @@
|
|
1 |
---
|
2 |
-
license: apache-2.0
|
3 |
language: ja
|
|
|
4 |
tags:
|
5 |
- audio
|
6 |
- automatic-speech-recognition
|
7 |
- hf-asr-leaderboard
|
|
|
|
|
8 |
widget:
|
9 |
- example_title: CommonVoice 8.0 (Test Split)
|
10 |
-
src:
|
11 |
-
https://huggingface.co/datasets/japanese-asr/ja_asr.common_voice_8_0/resolve/main/sample.flac
|
12 |
- example_title: JSUT Basic 5000
|
13 |
-
src:
|
14 |
-
https://huggingface.co/datasets/japanese-asr/ja_asr.jsut_basic5000/resolve/main/sample.flac
|
15 |
- example_title: ReazonSpeech (Test Split)
|
16 |
-
src:
|
17 |
-
https://huggingface.co/datasets/japanese-asr/ja_asr.reazonspeech_test/resolve/main/sample.flac
|
18 |
pipeline_tag: automatic-speech-recognition
|
19 |
-
metrics:
|
20 |
-
- wer
|
21 |
model-index:
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
---
|
61 |
|
62 |
# Kotoba-Whisper-v1.1
|
|
|
1 |
---
|
|
|
2 |
language: ja
|
3 |
+
license: apache-2.0
|
4 |
tags:
|
5 |
- audio
|
6 |
- automatic-speech-recognition
|
7 |
- hf-asr-leaderboard
|
8 |
+
metrics:
|
9 |
+
- wer
|
10 |
widget:
|
11 |
- example_title: CommonVoice 8.0 (Test Split)
|
12 |
+
src: https://huggingface.co/datasets/japanese-asr/ja_asr.common_voice_8_0/resolve/main/sample.flac
|
|
|
13 |
- example_title: JSUT Basic 5000
|
14 |
+
src: https://huggingface.co/datasets/japanese-asr/ja_asr.jsut_basic5000/resolve/main/sample.flac
|
|
|
15 |
- example_title: ReazonSpeech (Test Split)
|
16 |
+
src: https://huggingface.co/datasets/japanese-asr/ja_asr.reazonspeech_test/resolve/main/sample.flac
|
|
|
17 |
pipeline_tag: automatic-speech-recognition
|
|
|
|
|
18 |
model-index:
|
19 |
+
- name: kotoba-tech/kotoba-whisper-v1.0
|
20 |
+
results:
|
21 |
+
- task:
|
22 |
+
type: automatic-speech-recognition
|
23 |
+
dataset:
|
24 |
+
name: CommonVoice_8.0 (Japanese)
|
25 |
+
type: japanese-asr/ja_asr.common_voice_8_0
|
26 |
+
metrics:
|
27 |
+
- type: WER
|
28 |
+
value: 59.27
|
29 |
+
name: WER
|
30 |
+
- type: CER
|
31 |
+
value: 9.44
|
32 |
+
name: CER
|
33 |
+
- task:
|
34 |
+
type: automatic-speech-recognition
|
35 |
+
dataset:
|
36 |
+
name: ReazonSpeech (Test)
|
37 |
+
type: japanese-asr/ja_asr.reazonspeech_test
|
38 |
+
metrics:
|
39 |
+
- type: WER
|
40 |
+
value: 56.62
|
41 |
+
name: WER
|
42 |
+
- type: CER
|
43 |
+
value: 12.6
|
44 |
+
name: CER
|
45 |
+
- task:
|
46 |
+
type: automatic-speech-recognition
|
47 |
+
dataset:
|
48 |
+
name: JSUT Basic5000
|
49 |
+
type: japanese-asr/ja_asr.jsut_basic5000
|
50 |
+
metrics:
|
51 |
+
- type: WER
|
52 |
+
value: 64.36
|
53 |
+
name: WER
|
54 |
+
- type: CER
|
55 |
+
value: 8.48
|
56 |
+
name: CER
|
57 |
---
|
58 |
|
59 |
# Kotoba-Whisper-v1.1
|
config.json
CHANGED
@@ -54,7 +54,7 @@
|
|
54 |
"pad_token_id": 50256,
|
55 |
"scale_embedding": false,
|
56 |
"torch_dtype": "float32",
|
57 |
-
"transformers_version": "4.
|
58 |
"use_cache": true,
|
59 |
"use_weighted_layer_sum": false,
|
60 |
"vocab_size": 51866
|
|
|
54 |
"pad_token_id": 50256,
|
55 |
"scale_embedding": false,
|
56 |
"torch_dtype": "float32",
|
57 |
+
"transformers_version": "4.40.1",
|
58 |
"use_cache": true,
|
59 |
"use_weighted_layer_sum": false,
|
60 |
"vocab_size": 51866
|
generation_config.json
CHANGED
@@ -261,5 +261,5 @@
|
|
261 |
"transcribe": 50360,
|
262 |
"translate": 50359
|
263 |
},
|
264 |
-
"transformers_version": "4.
|
265 |
}
|
|
|
261 |
"transcribe": 50360,
|
262 |
"translate": 50359
|
263 |
},
|
264 |
+
"transformers_version": "4.40.1"
|
265 |
}
|
kotoba_whisper.py
CHANGED
@@ -24,7 +24,7 @@ class Punctuator:
|
|
24 |
def punctuate(self, pipeline_chunk: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
|
25 |
|
26 |
def validate_punctuation(raw: str, punctuated: str):
|
27 |
-
if 'unk' in punctuated:
|
28 |
return raw
|
29 |
if punctuated.count("。") > 1:
|
30 |
ind = punctuated.rfind("。")
|
|
|
24 |
def punctuate(self, pipeline_chunk: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
|
25 |
|
26 |
def validate_punctuation(raw: str, punctuated: str):
|
27 |
+
if 'unk' in punctuated.lower() or any(p in raw for p in self.ja_punctuations):
|
28 |
return raw
|
29 |
if punctuated.count("。") > 1:
|
30 |
ind = punctuated.rfind("。")
|