File size: 4,263 Bytes
ef11a50 5fb8a26 ef11a50 5fb8a26 ef11a50 5fb8a26 ef11a50 5fb8a26 ef11a50 5fb8a26 ef11a50 5fb8a26 |
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 76 77 78 79 80 81 82 83 84 85 86 87 |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- audiofolder
metrics:
- accuracy
- f1
- recall
- precision
model-index:
- name: wav2vec2-base-Speech_Emotion_Recognition
results: []
language:
- en
pipeline_tag: audio-classification
---
# wav2vec2-base-Speech_Emotion_Recognition
This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the audiofolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7264
- Accuracy: 0.7539
- Weighted f1: 0.7514
- Micro f1: 0.7539
- Macro f1: 0.7529
- Weighted recall: 0.7539
- Micro recall: 0.7539
- Macro recall: 0.7577
- Weighted precision: 0.7565
- Micro precision: 0.7539
- Macro precision: 0.7558
## Model description
This model predicts the emotion of the person speaking in the audio sample.
For more information on how it was created, check out the following link: https://github.com/DunnBC22/Vision_Audio_and_Multimodal_Projects/tree/main/Audio-Projects/Emotion%20Detection/Speech%20Emotion%20Detection
## Intended uses & limitations
This model is intended to demonstrate my ability to solve a complex problem using technology.
## Training and evaluation data
Dataset Source: https://www.kaggle.com/datasets/dmitrybabko/speech-emotion-recognition-en
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Weighted f1 | Micro f1 | Macro f1 | Weighted recall | Micro recall | Macro recall | Weighted precision | Micro precision | Macro precision |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-----------:|:--------:|:--------:|:---------------:|:------------:|:------------:|:------------------:|:---------------:|:---------------:|
| 1.5581 | 0.98 | 43 | 1.4046 | 0.4653 | 0.4080 | 0.4653 | 0.4174 | 0.4653 | 0.4653 | 0.4793 | 0.5008 | 0.4653 | 0.4974 |
| 1.5581 | 1.98 | 86 | 1.1566 | 0.5997 | 0.5836 | 0.5997 | 0.5871 | 0.5997 | 0.5997 | 0.6093 | 0.6248 | 0.5997 | 0.6209 |
| 1.5581 | 2.98 | 129 | 0.9733 | 0.6883 | 0.6845 | 0.6883 | 0.6860 | 0.6883 | 0.6883 | 0.6923 | 0.7012 | 0.6883 | 0.7009 |
| 1.5581 | 3.98 | 172 | 0.8313 | 0.7399 | 0.7392 | 0.7399 | 0.7409 | 0.7399 | 0.7399 | 0.7417 | 0.7415 | 0.7399 | 0.7432 |
| 1.5581 | 4.98 | 215 | 0.8708 | 0.7028 | 0.6963 | 0.7028 | 0.6970 | 0.7028 | 0.7028 | 0.7081 | 0.7148 | 0.7028 | 0.7114 |
| 1.5581 | 5.98 | 258 | 0.7969 | 0.7297 | 0.7267 | 0.7297 | 0.7277 | 0.7297 | 0.7297 | 0.7333 | 0.7393 | 0.7297 | 0.7382 |
| 1.5581 | 6.98 | 301 | 0.7349 | 0.7603 | 0.7613 | 0.7603 | 0.7631 | 0.7603 | 0.7603 | 0.7635 | 0.7699 | 0.7603 | 0.7702 |
| 1.5581 | 7.98 | 344 | 0.7714 | 0.7469 | 0.7444 | 0.7469 | 0.7456 | 0.7469 | 0.7469 | 0.7485 | 0.7554 | 0.7469 | 0.7563 |
| 1.5581 | 8.98 | 387 | 0.7183 | 0.7630 | 0.7615 | 0.7630 | 0.7631 | 0.7630 | 0.7630 | 0.7652 | 0.7626 | 0.7630 | 0.7637 |
| 1.5581 | 9.98 | 430 | 0.7264 | 0.7539 | 0.7514 | 0.7539 | 0.7529 | 0.7539 | 0.7539 | 0.7577 | 0.7565 | 0.7539 | 0.7558 |
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.0+cu118
- Datasets 2.11.0
- Tokenizers 0.13.3 |