File size: 2,774 Bytes
9a790f3
f13a285
9a790f3
 
f13a285
 
9a790f3
f13a285
 
9a790f3
f13a285
 
9a790f3
 
f13a285
9a790f3
f13a285
9a790f3
f13a285
9a790f3
f13a285
9a790f3
f13a285
 
 
9a790f3
f13a285
9a790f3
f13a285
9a790f3
f13a285
 
 
9a790f3
 
 
 
 
 
f13a285
 
 
 
 
 
 
 
 
 
 
9a790f3
 
 
f13a285
 
 
 
 
 
 
 
 
 
 
 
9a790f3
f13a285
9a790f3
f13a285
9a790f3
f13a285
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
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
63
64
65
66
67
68
69
70
71
72
73
74
75
---
language: id
license: apache-2.0
tags:
    - audio-classification
    - generated_from_trainer
metrics:
    - accuracy
    - f1
model-index:
    - name: wav2vec2-xls-r-adult-child-id-cls
      results: []
---

# Wav2Vec2 XLS-R Adult/Child Indonesian Speech Classifier

Wav2Vec2 XLS-R Adult/Child Indonesian Speech Classifier is an audio classification model based on the [XLS-R](https://arxiv.org/abs/2111.09296) architecture. This model is a fine-tuned version of [wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on a private adult/child Indonesian speech classification dataset.

This model was trained using HuggingFace's PyTorch framework. All training was done on a Tesla P100, provided by Kaggle. Training metrics were logged via Tensorboard.

## Model

| Model                               | #params | Arch. | Training/Validation data (text)                      |
| ----------------------------------- | ------- | ----- | ---------------------------------------------------- |
| `wav2vec2-xls-r-adult-child-id-cls` | 300M    | XLS-R | Adult/Child Indonesian Speech Classification Dataset |

## Evaluation Results

The model achieves the following results on evaluation:

| Dataset                                      | Loss   | Accuracy | F1     |
| -------------------------------------------- | ------ | -------- | ------ |
| Adult/Child Indonesian Speech Classification | 0.1970 | 93.38%   | 0.9307 |

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:

-   `learning_rate`: 3e-05
-   `train_batch_size`: 8
-   `eval_batch_size`: 8
-   `seed`: 42
-   `gradient_accumulation_steps`: 4
-   `total_train_batch_size`: 32
-   `optimizer`: Adam with `betas=(0.9,0.999)` and `epsilon=1e-08`
-   `lr_scheduler_type`: linear
-   `lr_scheduler_warmup_ratio`: 0.1
-   `num_epochs`: 4

### Training results

| Training Loss | Epoch | Step | Validation Loss | Accuracy |   F1   |
| :-----------: | :---: | :--: | :-------------: | :------: | :----: |
|     0.336     |  1.0  | 305  |     0.3146      |  0.8845  | 0.8698 |
|    0.2345     |  2.0  | 610  |     0.2140      |  0.9251  | 0.9202 |
|    0.3215     |  3.0  | 915  |     0.2038      |  0.9315  | 0.9286 |
|    0.2059     |  4.0  | 1220 |     0.1970      |  0.9338  | 0.9307 |

## Disclaimer

Do consider the biases which came from pre-training datasets that may be carried over into the results of this model.

## Authors

Wav2Vec2 XLS-R Adult/Child Indonesian Speech Classifier was trained and evaluated by [Ananto Joyoadikusumo](https://anantoj.github.io/). All computation and development are done on Kaggle.

## Framework versions

-   Transformers 4.18.0
-   Pytorch 1.11.0+cu102
-   Datasets 2.2.0
-   Tokenizers 0.12.1