|
--- |
|
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 |
