Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,16 +1,60 @@
|
|
1 |
import torch
|
2 |
-
import
|
3 |
-
|
4 |
-
from transformers import WhisperForConditionalGeneration, WhisperProcessor
|
5 |
-
from transformers.models.whisper.tokenization_whisper import LANGUAGES
|
6 |
from transformers.pipelines.audio_utils import ffmpeg_read
|
7 |
-
|
8 |
import gradio as gr
|
9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
|
11 |
device = "cuda" if torch.cuda.is_available() else "CPU"
|
12 |
|
13 |
-
model_ckpt = "
|
14 |
model = AutoModelForSequenceClassification.from_pretrained(model_ckpt)
|
15 |
tokenizer = AutoTokenizer.from_pretrained(model_ckpt)
|
16 |
|
|
|
1 |
import torch
|
2 |
+
from transformers import AutoModelForSequenceClassification, AutoTokenizer
|
|
|
|
|
|
|
3 |
from transformers.pipelines.audio_utils import ffmpeg_read
|
|
|
4 |
import gradio as gr
|
5 |
|
6 |
+
LANGUANGE_MAP = {
|
7 |
+
0: 'Arabic',
|
8 |
+
1: 'Basque',
|
9 |
+
2: 'Breton',
|
10 |
+
3: 'Catalan',
|
11 |
+
4: 'Chinese_China',
|
12 |
+
5: 'Chinese_Hongkong',
|
13 |
+
6: 'Chinese_Taiwan',
|
14 |
+
7: 'Chuvash',
|
15 |
+
8: 'Czech',
|
16 |
+
9: 'Dhivehi',
|
17 |
+
10: 'Dutch',
|
18 |
+
11: 'English',
|
19 |
+
12: 'Esperanto',
|
20 |
+
13: 'Estonian',
|
21 |
+
14: 'French',
|
22 |
+
15: 'Frisian',
|
23 |
+
16: 'Georgian',
|
24 |
+
17: 'German',
|
25 |
+
18: 'Greek',
|
26 |
+
19: 'Hakha_Chin',
|
27 |
+
20: 'Indonesian',
|
28 |
+
21: 'Interlingua',
|
29 |
+
22: 'Italian',
|
30 |
+
23: 'Japanese',
|
31 |
+
24: 'Kabyle',
|
32 |
+
25: 'Kinyarwanda',
|
33 |
+
26: 'Kyrgyz',
|
34 |
+
27: 'Latvian',
|
35 |
+
28: 'Maltese',
|
36 |
+
29: 'Mongolian',
|
37 |
+
30: 'Persian',
|
38 |
+
31: 'Polish',
|
39 |
+
32: 'Portuguese',
|
40 |
+
33: 'Romanian',
|
41 |
+
34: 'Romansh_Sursilvan',
|
42 |
+
35: 'Russian',
|
43 |
+
36: 'Sakha',
|
44 |
+
37: 'Slovenian',
|
45 |
+
38: 'Spanish',
|
46 |
+
39: 'Swedish',
|
47 |
+
40: 'Tamil',
|
48 |
+
41: 'Tatar',
|
49 |
+
42: 'Turkish',
|
50 |
+
43: 'Ukranian',
|
51 |
+
44: 'Welsh'
|
52 |
+
}
|
53 |
+
|
54 |
|
55 |
device = "cuda" if torch.cuda.is_available() else "CPU"
|
56 |
|
57 |
+
model_ckpt = "barto17/language-detection-fine-tuned-on-xlm-roberta-base"
|
58 |
model = AutoModelForSequenceClassification.from_pretrained(model_ckpt)
|
59 |
tokenizer = AutoTokenizer.from_pretrained(model_ckpt)
|
60 |
|