Hugging Face's logo

mozilla-ai
/
whisper-large-v3-turbo-fi

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Finnish
whisper
Eval Results
Model card Files
xet
Metrics Community
kostissz commited on
Commit
08a3388
·
verified ·
1 Parent(s): 3ffe593

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +32 -157
README.md CHANGED
@@ -1,166 +1,41 @@
1
  ---
2
- library_name: transformers
3
- license: mit
4
  base_model: openai/whisper-large-v3-turbo
5
- tags:
6
- - generated_from_trainer
7
- metrics:
8
- - wer
 
9
  model-index:
10
- - name: whisper-large-v3-turbo-fi
11
- results: []
 
 
 
 
 
 
 
 
 
12
  ---
13
 
14
- <!-- This model card has been generated automatically according to the information the Trainer had access to. You
15
- should probably proofread and complete it, then remove this comment. -->
16
-
17
- # whisper-large-v3-turbo-fi
18
-
19
- This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on the None dataset.
20
- It achieves the following results on the evaluation set:
21
- - Loss: 0.1720
22
- - Model Preparation Time: 0.0039
23
- - Wer Ortho: 15.8841
24
- - Wer: 11.9718
25
- - Cer Ortho: 3.1186
26
- - Cer: 2.2547
27
-
28
- ## Model description
29
-
30
- More information needed
31
-
32
- ## Intended uses & limitations
33
-
34
- More information needed
35
-
36
- ## Training and evaluation data
37
-
38
- More information needed
39
-
40
- ## Training procedure
41
-
42
- ### Training hyperparameters
43
-
44
- The following hyperparameters were used during training:
45
- - learning_rate: 1e-05
46
- - train_batch_size: 64
47
- - eval_batch_size: 8
48
- - seed: 42
49
- - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
50
- - lr_scheduler_type: linear
51
- - lr_scheduler_warmup_steps: 50
52
- - training_steps: 5000
53
- - mixed_precision_training: Native AMP
54
-
55
- ### Training results
56
 
57
- | Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Wer Ortho | Wer | Cer Ortho | Cer |
58
- |:-------------:|:--------:|:----:|:---------------:|:----------------------:|:---------:|:-------:|:---------:|:------:|
59
- | 0.6065 | 1.5152 | 50 | 0.1720 | 0.0039 | 15.8841 | 11.9718 | 3.1186 | 2.2547 |
60
- | 0.0722 | 3.0303 | 100 | 0.1870 | 0.0039 | 15.8929 | 12.6232 | 2.6462 | 2.1404 |
61
- | 0.0283 | 4.5455 | 150 | 0.2010 | 0.0039 | 15.6902 | 12.4120 | 2.6704 | 2.1664 |
62
- | 0.018 | 6.0606 | 200 | 0.2224 | 0.0039 | 16.1748 | 12.9401 | 2.7827 | 2.2739 |
63
- | 0.0113 | 7.5758 | 250 | 0.2353 | 0.0039 | 16.6946 | 13.5123 | 2.8697 | 2.3508 |
64
- | 0.0091 | 9.0909 | 300 | 0.2553 | 0.0039 | 18.4213 | 15.2465 | 3.2232 | 2.7162 |
65
- | 0.0082 | 10.6061 | 350 | 0.2486 | 0.0039 | 17.6196 | 14.0933 | 3.0756 | 2.5228 |
66
- | 0.0086 | 12.1212 | 400 | 0.2547 | 0.0039 | 18.1041 | 14.8415 | 3.1296 | 2.6280 |
67
- | 0.0068 | 13.6364 | 450 | 0.2784 | 0.0039 | 18.5358 | 15.2905 | 3.3091 | 2.7943 |
68
- | 0.0061 | 15.1515 | 500 | 0.2723 | 0.0039 | 17.6989 | 14.4014 | 3.1362 | 2.6076 |
69
- | 0.005 | 16.6667 | 550 | 0.2642 | 0.0039 | 17.2584 | 14.0317 | 2.9523 | 2.4515 |
70
- | 0.004 | 18.1818 | 600 | 0.2766 | 0.0039 | 17.8222 | 14.6655 | 3.1857 | 2.6981 |
71
- | 0.003 | 19.6970 | 650 | 0.2777 | 0.0039 | 18.0953 | 14.9472 | 3.1142 | 2.6291 |
72
- | 0.0038 | 21.2121 | 700 | 0.2823 | 0.0039 | 18.0689 | 14.8504 | 3.1329 | 2.6257 |
73
- | 0.005 | 22.7273 | 750 | 0.2793 | 0.0039 | 17.8663 | 14.7007 | 3.1505 | 2.6472 |
74
- | 0.0031 | 24.2424 | 800 | 0.2986 | 0.0039 | 18.0513 | 14.8151 | 3.1362 | 2.6337 |
75
- | 0.0021 | 25.7576 | 850 | 0.2881 | 0.0039 | 18.7649 | 15.6514 | 3.2793 | 2.8000 |
76
- | 0.0033 | 27.2727 | 900 | 0.3008 | 0.0039 | 18.8706 | 15.7835 | 3.4126 | 2.9346 |
77
- | 0.0046 | 28.7879 | 950 | 0.3031 | 0.0039 | 18.7384 | 15.1937 | 3.2837 | 2.7445 |
78
- | 0.002 | 30.3030 | 1000 | 0.3213 | 0.0039 | 18.7472 | 15.5546 | 3.2904 | 2.7977 |
79
- | 0.0008 | 31.8182 | 1050 | 0.2988 | 0.0039 | 17.4786 | 14.2958 | 2.9853 | 2.4979 |
80
- | 0.0002 | 33.3333 | 1100 | 0.3072 | 0.0039 | 17.1703 | 13.9701 | 2.9204 | 2.4357 |
81
- | 0.0003 | 34.8485 | 1150 | 0.3102 | 0.0039 | 17.3465 | 14.1725 | 2.9435 | 2.4572 |
82
- | 0.0001 | 36.3636 | 1200 | 0.3126 | 0.0039 | 17.2936 | 14.1021 | 2.9049 | 2.4232 |
83
- | 0.0002 | 37.8788 | 1250 | 0.3136 | 0.0039 | 17.2232 | 14.0493 | 2.9060 | 2.4278 |
84
- | 0.0002 | 39.3939 | 1300 | 0.3153 | 0.0039 | 17.0205 | 13.8644 | 2.8675 | 2.3916 |
85
- | 0.0001 | 40.9091 | 1350 | 0.3173 | 0.0039 | 17.1615 | 13.9613 | 2.8796 | 2.4029 |
86
- | 0.0001 | 42.4242 | 1400 | 0.3185 | 0.0039 | 17.1174 | 13.9261 | 2.8785 | 2.4017 |
87
- | 0.0001 | 43.9394 | 1450 | 0.3200 | 0.0039 | 17.0646 | 13.8996 | 2.8719 | 2.3984 |
88
- | 0.0001 | 45.4545 | 1500 | 0.3213 | 0.0039 | 17.0293 | 13.8820 | 2.8675 | 2.3950 |
89
- | 0.0001 | 46.9697 | 1550 | 0.3224 | 0.0039 | 17.0381 | 13.8996 | 2.8686 | 2.3961 |
90
- | 0.0001 | 48.4848 | 1600 | 0.3236 | 0.0039 | 16.9589 | 13.8204 | 2.8576 | 2.3848 |
91
- | 0.0001 | 50.0 | 1650 | 0.3246 | 0.0039 | 16.9765 | 13.8292 | 2.8532 | 2.3825 |
92
- | 0.0001 | 51.5152 | 1700 | 0.3257 | 0.0039 | 16.9677 | 13.8204 | 2.8664 | 2.3927 |
93
- | 0.0001 | 53.0303 | 1750 | 0.3268 | 0.0039 | 16.9412 | 13.7940 | 2.8653 | 2.3927 |
94
- | 0.0001 | 54.5455 | 1800 | 0.3276 | 0.0039 | 16.9236 | 13.7852 | 2.8587 | 2.3870 |
95
- | 0.0001 | 56.0606 | 1850 | 0.3287 | 0.0039 | 16.8972 | 13.7676 | 2.8521 | 2.3791 |
96
- | 0.0001 | 57.5758 | 1900 | 0.3296 | 0.0039 | 16.8972 | 13.7676 | 2.8565 | 2.3836 |
97
- | 0.0001 | 59.0909 | 1950 | 0.3305 | 0.0039 | 16.9324 | 13.7940 | 2.8730 | 2.3995 |
98
- | 0.0001 | 60.6061 | 2000 | 0.3314 | 0.0039 | 16.9324 | 13.7940 | 2.8708 | 2.3972 |
99
- | 0.0001 | 62.1212 | 2050 | 0.3323 | 0.0039 | 16.9236 | 13.7764 | 2.8686 | 2.3938 |
100
- | 0.0001 | 63.6364 | 2100 | 0.3330 | 0.0039 | 16.9148 | 13.7940 | 2.8708 | 2.3972 |
101
- | 0.0001 | 65.1515 | 2150 | 0.3339 | 0.0039 | 16.9412 | 13.8204 | 2.8697 | 2.3972 |
102
- | 0.0001 | 66.6667 | 2200 | 0.3348 | 0.0039 | 16.9500 | 13.8204 | 2.8686 | 2.3961 |
103
- | 0.0001 | 68.1818 | 2250 | 0.3355 | 0.0039 | 16.8884 | 13.7676 | 2.8664 | 2.3950 |
104
- | 0.0001 | 69.6970 | 2300 | 0.3363 | 0.0039 | 16.8796 | 13.7676 | 2.8609 | 2.3904 |
105
- | 0.0001 | 71.2121 | 2350 | 0.3370 | 0.0039 | 16.8796 | 13.7500 | 2.8642 | 2.3927 |
106
- | 0.0001 | 72.7273 | 2400 | 0.3379 | 0.0039 | 16.8796 | 13.7676 | 2.8620 | 2.3916 |
107
- | 0.0 | 74.2424 | 2450 | 0.3385 | 0.0039 | 16.8972 | 13.7940 | 2.8653 | 2.3961 |
108
- | 0.0 | 75.7576 | 2500 | 0.3392 | 0.0039 | 16.9060 | 13.8028 | 2.8697 | 2.4006 |
109
- | 0.0 | 77.2727 | 2550 | 0.3399 | 0.0039 | 16.8884 | 13.7852 | 2.8697 | 2.3995 |
110
- | 0.0 | 78.7879 | 2600 | 0.3407 | 0.0039 | 16.8884 | 13.7940 | 2.8719 | 2.4017 |
111
- | 0.0 | 80.3030 | 2650 | 0.3413 | 0.0039 | 16.8972 | 13.7940 | 2.8708 | 2.3995 |
112
- | 0.0 | 81.8182 | 2700 | 0.3422 | 0.0039 | 16.8796 | 13.7764 | 2.8675 | 2.3972 |
113
- | 0.0 | 83.3333 | 2750 | 0.3427 | 0.0039 | 16.8179 | 13.7148 | 2.8554 | 2.3848 |
114
- | 0.0 | 84.8485 | 2800 | 0.3433 | 0.0039 | 16.8267 | 13.7236 | 2.8521 | 2.3825 |
115
- | 0.0 | 86.3636 | 2850 | 0.3440 | 0.0039 | 16.8355 | 13.7324 | 2.8620 | 2.3927 |
116
- | 0.0 | 87.8788 | 2900 | 0.3447 | 0.0039 | 16.7827 | 13.6796 | 2.8554 | 2.3859 |
117
- | 0.0 | 89.3939 | 2950 | 0.3453 | 0.0039 | 16.7739 | 13.6620 | 2.8510 | 2.3803 |
118
- | 0.0 | 90.9091 | 3000 | 0.3459 | 0.0039 | 16.7739 | 13.6620 | 2.8510 | 2.3803 |
119
- | 0.0 | 92.4242 | 3050 | 0.3465 | 0.0039 | 16.7915 | 13.6620 | 2.8532 | 2.3814 |
120
- | 0.0 | 93.9394 | 3100 | 0.3471 | 0.0039 | 16.7827 | 13.6620 | 2.8543 | 2.3848 |
121
- | 0.0 | 95.4545 | 3150 | 0.3476 | 0.0039 | 16.8003 | 13.6708 | 2.8554 | 2.3848 |
122
- | 0.0 | 96.9697 | 3200 | 0.3482 | 0.0039 | 16.7562 | 13.6268 | 2.8488 | 2.3780 |
123
- | 0.0 | 98.4848 | 3250 | 0.3487 | 0.0039 | 16.7650 | 13.6356 | 2.8554 | 2.3848 |
124
- | 0.0 | 100.0 | 3300 | 0.3493 | 0.0039 | 16.7210 | 13.5915 | 2.8510 | 2.3803 |
125
- | 0.0 | 101.5152 | 3350 | 0.3499 | 0.0039 | 16.6946 | 13.5739 | 2.8477 | 2.3780 |
126
- | 0.0 | 103.0303 | 3400 | 0.3504 | 0.0039 | 16.6946 | 13.5827 | 2.8477 | 2.3780 |
127
- | 0.0 | 104.5455 | 3450 | 0.3508 | 0.0039 | 16.6858 | 13.5739 | 2.8466 | 2.3769 |
128
- | 0.0 | 106.0606 | 3500 | 0.3513 | 0.0039 | 16.6858 | 13.5739 | 2.8466 | 2.3769 |
129
- | 0.0 | 107.5758 | 3550 | 0.3518 | 0.0039 | 16.6681 | 13.5563 | 2.8422 | 2.3723 |
130
- | 0.0 | 109.0909 | 3600 | 0.3524 | 0.0039 | 16.6593 | 13.5475 | 2.8411 | 2.3712 |
131
- | 0.0 | 110.6061 | 3650 | 0.3528 | 0.0039 | 16.6329 | 13.5211 | 2.8378 | 2.3678 |
132
- | 0.0 | 112.1212 | 3700 | 0.3532 | 0.0039 | 16.7122 | 13.6004 | 2.8422 | 2.3723 |
133
- | 0.0 | 113.6364 | 3750 | 0.3538 | 0.0039 | 16.6769 | 13.5651 | 2.8400 | 2.3701 |
134
- | 0.0 | 115.1515 | 3800 | 0.3541 | 0.0039 | 16.6946 | 13.5827 | 2.8422 | 2.3723 |
135
- | 0.0 | 116.6667 | 3850 | 0.3545 | 0.0039 | 16.7034 | 13.5915 | 2.8411 | 2.3723 |
136
- | 0.0 | 118.1818 | 3900 | 0.3550 | 0.0039 | 16.7034 | 13.5915 | 2.8422 | 2.3746 |
137
- | 0.0 | 119.6970 | 3950 | 0.3554 | 0.0039 | 16.7122 | 13.6004 | 2.8411 | 2.3735 |
138
- | 0.0 | 121.2121 | 4000 | 0.3557 | 0.0039 | 16.7122 | 13.6004 | 2.8389 | 2.3712 |
139
- | 0.0 | 122.7273 | 4050 | 0.3561 | 0.0039 | 16.7122 | 13.6004 | 2.8389 | 2.3712 |
140
- | 0.0 | 124.2424 | 4100 | 0.3565 | 0.0039 | 16.7034 | 13.5915 | 2.8411 | 2.3735 |
141
- | 0.0 | 125.7576 | 4150 | 0.3568 | 0.0039 | 16.7210 | 13.6092 | 2.8400 | 2.3723 |
142
- | 0.0 | 127.2727 | 4200 | 0.3571 | 0.0039 | 16.7298 | 13.6092 | 2.8400 | 2.3723 |
143
- | 0.0 | 128.7879 | 4250 | 0.3574 | 0.0039 | 16.6858 | 13.5651 | 2.8356 | 2.3678 |
144
- | 0.0 | 130.3030 | 4300 | 0.3577 | 0.0039 | 16.7034 | 13.5915 | 2.8356 | 2.3689 |
145
- | 0.0 | 131.8182 | 4350 | 0.3579 | 0.0039 | 16.6593 | 13.5387 | 2.8290 | 2.3622 |
146
- | 0.0 | 133.3333 | 4400 | 0.3582 | 0.0039 | 16.6593 | 13.5387 | 2.8312 | 2.3644 |
147
- | 0.0 | 134.8485 | 4450 | 0.3584 | 0.0039 | 16.6769 | 13.5563 | 2.8334 | 2.3667 |
148
- | 0.0 | 136.3636 | 4500 | 0.3586 | 0.0039 | 16.6505 | 13.5299 | 2.8290 | 2.3622 |
149
- | 0.0 | 137.8788 | 4550 | 0.3588 | 0.0039 | 16.6417 | 13.5211 | 2.8268 | 2.3599 |
150
- | 0.0 | 139.3939 | 4600 | 0.3590 | 0.0039 | 16.6417 | 13.5211 | 2.8312 | 2.3644 |
151
- | 0.0 | 140.9091 | 4650 | 0.3592 | 0.0039 | 16.6505 | 13.5299 | 2.8301 | 2.3633 |
152
- | 0.0 | 142.4242 | 4700 | 0.3593 | 0.0039 | 16.6241 | 13.5035 | 2.8256 | 2.3588 |
153
- | 0.0 | 143.9394 | 4750 | 0.3595 | 0.0039 | 16.6241 | 13.5035 | 2.8279 | 2.3610 |
154
- | 0.0 | 145.4545 | 4800 | 0.3596 | 0.0039 | 16.6241 | 13.5035 | 2.8279 | 2.3610 |
155
- | 0.0 | 146.9697 | 4850 | 0.3596 | 0.0039 | 16.6241 | 13.5035 | 2.8279 | 2.3610 |
156
- | 0.0 | 148.4848 | 4900 | 0.3597 | 0.0039 | 16.6241 | 13.5035 | 2.8268 | 2.3599 |
157
- | 0.0 | 150.0 | 4950 | 0.3597 | 0.0039 | 16.6241 | 13.5035 | 2.8290 | 2.3622 |
158
- | 0.0 | 151.5152 | 5000 | 0.3597 | 0.0039 | 16.6241 | 13.5035 | 2.8268 | 2.3599 |
159
 
 
160
 
161
- ### Framework versions
 
 
 
 
 
162
 
163
- - Transformers 4.52.4
164
- - Pytorch 2.7.1+cu126
165
- - Datasets 3.6.0
166
- - Tokenizers 0.21.2
 
 
 
1
  ---
 
 
2
  base_model: openai/whisper-large-v3-turbo
3
+ datasets:
4
+ - fi
5
+ language: fi
6
+ library_name: transformers
7
+ license: apache-2.0
8
  model-index:
9
+ - name: Finetuned openai/whisper-large-v3-turbo on Finnish
10
+ results:
11
+ - task:
12
+ type: automatic-speech-recognition
13
+ name: Speech-to-Text
14
+ dataset:
15
+ name: Common Voice (Finnish)
16
+ type: common_voice
17
+ metrics:
18
+ - type: wer
19
+ value: 11.972
20
  ---
21
 
22
+ # Finetuned openai/whisper-large-v3-turbo on 2093 Finnish training audio samples from /home/kostis/cv-corpus-22.0-2025-06-20/fi.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23
 
24
+ This model was created from the Mozilla.ai Blueprint:
25
+ [speech-to-text-finetune](https://github.com/mozilla-ai/speech-to-text-finetune).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
26
 
27
+ ## Evaluation results on 1806 audio samples of Finnish:
28
 
29
+ ### Baseline model (before finetuning) on Finnish
30
+ - Word Error Rate (Normalized): 16.576
31
+ - Word Error Rate (Orthographic): 20.298
32
+ - Character Error Rate (Normalized): 3.808
33
+ - Character Error Rate (Orthographic): 4.498
34
+ - Loss: 2.112
35
 
36
+ ### Finetuned model (after finetuning) on Finnish
37
+ - Word Error Rate (Normalized): 11.972
38
+ - Word Error Rate (Orthographic): 15.884
39
+ - Character Error Rate (Normalized): 2.255
40
+ - Character Error Rate (Orthographic): 3.119
41
+ - Loss: 0.172