Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,138 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
datasets:
|
3 |
+
- thesven/bengali-ai-train-set-tiny
|
4 |
+
language:
|
5 |
+
- bn
|
6 |
+
metrics:
|
7 |
+
- wer
|
8 |
+
---
|
9 |
+
# thesven/whisper-tiny-bn-v2
|
10 |
+
|
11 |
+
|
12 |
+
|
13 |
+
## Model Description
|
14 |
+
|
15 |
+
|
16 |
+
This is the `openai/whisper-tiny` model fine-tuned on the OOD-Speech Bengali dataset. The fine-tuning was performed to enhance the model's performance on Bengali speech recognition tasks, especially for out-of-distribution samples.
|
17 |
+
|
18 |
+
|
19 |
+
|
20 |
+
## Model Details
|
21 |
+
|
22 |
+
|
23 |
+
|
24 |
+
- **Original Model:** [openai/whisper-tiny on Hugging Face](https://huggingface.co/openai/whisper-tiny)
|
25 |
+
|
26 |
+
- **Fine-tuned Model:** [thesven/whisper-tiny-bn-thesven on Hugging Face](https://huggingface.co/thesven/whisper-tiny-bn-thesven)
|
27 |
+
|
28 |
+
- **Model Architecture:** Whisper ASR
|
29 |
+
|
30 |
+
|
31 |
+
|
32 |
+
## Intended Use
|
33 |
+
|
34 |
+
|
35 |
+
|
36 |
+
This model is intended for automatic speech recognition (ASR) tasks in the Bengali language, especially for scenarios that involve out-of-distribution samples.
|
37 |
+
|
38 |
+
|
39 |
+
## Training Data
|
40 |
+
|
41 |
+
|
42 |
+
|
43 |
+
The model was fine-tuned using the OOD-Speech Bengali dataset:
|
44 |
+
|
45 |
+
|
46 |
+
|
47 |
+
- **Training Samples:** 10,000
|
48 |
+
|
49 |
+
- **Validation Samples:** 1,000
|
50 |
+
|
51 |
+
- **Dataset Source:** [OOD-Speech: A Large Bengali Speech Recognition Dataset for Out-of-Distribution Benchmarking](https://arxiv.org/abs/2305.09688)
|
52 |
+
|
53 |
+
|
54 |
+
|
55 |
+
## Fine-tuning Procedure
|
56 |
+
|
57 |
+
|
58 |
+
|
59 |
+
The model was fine-tuned using the OOD-Speech Bengali dataset with [specific hyperparameters, optimizer, learning rate, epochs, etc. if available].
|
60 |
+
|
61 |
+
|
62 |
+
|
63 |
+
## Licensing
|
64 |
+
|
65 |
+
|
66 |
+
|
67 |
+
This model is released under the MIT License. You are free to use, modify, distribute, or sell it under the terms of the license.
|
68 |
+
|
69 |
+
|
70 |
+
|
71 |
+
```
|
72 |
+
|
73 |
+
MIT License
|
74 |
+
|
75 |
+
|
76 |
+
|
77 |
+
Copyright (c) 2023 thesven
|
78 |
+
|
79 |
+
|
80 |
+
|
81 |
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
82 |
+
|
83 |
+
of this software and associated documentation files (the "Software"), to deal
|
84 |
+
|
85 |
+
in the Software without restriction, including without limitation the rights
|
86 |
+
|
87 |
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
88 |
+
|
89 |
+
copies of the Software, and to permit persons to whom the Software is
|
90 |
+
|
91 |
+
furnished to do so, subject to the following conditions:
|
92 |
+
|
93 |
+
|
94 |
+
|
95 |
+
The above copyright notice and this permission notice shall be included in all
|
96 |
+
|
97 |
+
copies or substantial portions of the Software.
|
98 |
+
|
99 |
+
|
100 |
+
|
101 |
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
102 |
+
|
103 |
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
104 |
+
|
105 |
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
106 |
+
|
107 |
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
108 |
+
|
109 |
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
110 |
+
|
111 |
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
112 |
+
|
113 |
+
SOFTWARE.
|
114 |
+
|
115 |
+
```
|
116 |
+
|
117 |
+
|
118 |
+
|
119 |
+
|
120 |
+
## Citation
|
121 |
+
|
122 |
+
|
123 |
+
|
124 |
+
```
|
125 |
+
|
126 |
+
@article{OOD-Speech2023,
|
127 |
+
|
128 |
+
title={OOD-Speech: A Large Bengali Speech Recognition Dataset for Out-of-Distribution Benchmarking},
|
129 |
+
|
130 |
+
author={Authors of the OOD-Speech paper},
|
131 |
+
|
132 |
+
journal={arXiv preprint arXiv:2305.09688},
|
133 |
+
|
134 |
+
year={2023}
|
135 |
+
|
136 |
+
}
|
137 |
+
|
138 |
+
```
|