Audio Classification
Transformers
PyTorch
TensorBoard
Safetensors
English
audio-spectrogram-transformer
Generated from Trainer
Instructions to use bookbot/distil-ast-audioset with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use bookbot/distil-ast-audioset with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("audio-classification", model="bookbot/distil-ast-audioset")# pip install -U transformers accelerate # Load model directly from transformers import AutoFeatureExtractor, AutoModelForAudioClassification extractor = AutoFeatureExtractor.from_pretrained("bookbot/distil-ast-audioset") model = AutoModelForAudioClassification.from_pretrained("bookbot/distil-ast-audioset", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,75 +1,77 @@
|
|
| 1 |
---
|
| 2 |
-
|
|
|
|
| 3 |
tags:
|
| 4 |
-
- audio-classification
|
| 5 |
-
- generated_from_trainer
|
| 6 |
metrics:
|
| 7 |
-
-
|
| 8 |
-
-
|
| 9 |
-
model-index:
|
| 10 |
-
- name: distil-ast-audioset-2
|
| 11 |
-
results: []
|
| 12 |
---
|
| 13 |
|
| 14 |
-
|
| 15 |
-
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
|
| 17 |
-
|
| 18 |
|
| 19 |
-
This model
|
| 20 |
-
It achieves the following results on the evaluation set:
|
| 21 |
-
- Loss: 0.3063
|
| 22 |
-
- F1: 0.4876
|
| 23 |
-
- Roc Auc: 0.7140
|
| 24 |
-
- Accuracy: 0.0714
|
| 25 |
-
- Map: 0.4743
|
| 26 |
|
| 27 |
-
## Model
|
| 28 |
|
| 29 |
-
|
|
|
|
|
|
|
| 30 |
|
| 31 |
-
##
|
| 32 |
|
| 33 |
-
|
| 34 |
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
|
|
|
| 38 |
|
| 39 |
## Training procedure
|
| 40 |
|
| 41 |
### Training hyperparameters
|
| 42 |
|
| 43 |
The following hyperparameters were used during training:
|
| 44 |
-
|
| 45 |
-
-
|
| 46 |
-
-
|
| 47 |
-
-
|
| 48 |
-
-
|
| 49 |
-
-
|
| 50 |
-
-
|
| 51 |
-
-
|
| 52 |
-
-
|
| 53 |
-
-
|
| 54 |
-
-
|
|
|
|
| 55 |
|
| 56 |
### Training results
|
| 57 |
|
| 58 |
-
| Training Loss | Epoch | Step
|
| 59 |
-
|:-----------
|
| 60 |
-
|
|
| 61 |
-
|
|
| 62 |
-
|
|
| 63 |
-
|
|
| 64 |
-
|
|
| 65 |
-
|
|
| 66 |
-
|
|
| 67 |
-
|
|
| 68 |
-
|
|
| 69 |
-
|
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73 |
|
| 74 |
- Transformers 4.27.0.dev0
|
| 75 |
- Pytorch 1.13.1+cu117
|
|
|
|
| 1 |
---
|
| 2 |
+
language: en
|
| 3 |
+
license: apache-2.0
|
| 4 |
tags:
|
| 5 |
+
- audio-classification
|
| 6 |
+
- generated_from_trainer
|
| 7 |
metrics:
|
| 8 |
+
- accuracy
|
| 9 |
+
- f1
|
|
|
|
|
|
|
|
|
|
| 10 |
---
|
| 11 |
|
| 12 |
+
# Distil Audio Spectrogram Transformer AudioSet
|
|
|
|
| 13 |
|
| 14 |
+
Distil Audio Spectrogram Transformer AudioSet is an audio classification model based on the [Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) architecture. This model is a distilled version of [MIT/ast-finetuned-audioset-10-10-0.4593](https://huggingface.co/MIT/ast-finetuned-audioset-10-10-0.4593) on the [AudioSet](https://research.google.com/audioset/download.html) dataset.
|
| 15 |
|
| 16 |
+
This model was trained using HuggingFace's PyTorch framework. All training was done on a Google Cloud Engine VM with a Tesla A100 GPU. All necessary scripts used for training could be found in the [Files and versions](https://huggingface.co/bookbot/distil-ast-audioset/tree/main) tab, as well as the [Training metrics](https://huggingface.co/bookbot/distil-ast-audioset/tensorboard) logged via Tensorboard.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
+
## Model
|
| 19 |
|
| 20 |
+
| Model | #params | Arch. | Training/Validation data |
|
| 21 |
+
| --------------------- | ------- | ----------------------------- | ------------------------ |
|
| 22 |
+
| `distil-ast-audioset` | 44M | Audio Spectrogram Transformer | AudioSet |
|
| 23 |
|
| 24 |
+
## Evaluation Results
|
| 25 |
|
| 26 |
+
The model achieves the following results on evaluation:
|
| 27 |
|
| 28 |
+
| Model | F1 | Roc Auc | Accuracy | mAP |
|
| 29 |
+
| ------------------- | ------ | ------- | -------- | ------ |
|
| 30 |
+
| Distil-AST AudioSet | 0.4876 | 0.7140 | 0.0714 | 0.4743 |
|
| 31 |
+
| AST AudioSet | 0.4989 | 0.6905 | 0.1247 | 0.5603 |
|
| 32 |
|
| 33 |
## Training procedure
|
| 34 |
|
| 35 |
### Training hyperparameters
|
| 36 |
|
| 37 |
The following hyperparameters were used during training:
|
| 38 |
+
|
| 39 |
+
- `learning_rate`: 3e-05
|
| 40 |
+
- `train_batch_size`: 32
|
| 41 |
+
- `eval_batch_size`: 32
|
| 42 |
+
- `seed`: 0
|
| 43 |
+
- `gradient_accumulation_steps`: 4
|
| 44 |
+
- `total_train_batch_size`: 128
|
| 45 |
+
- `optimizer`: Adam with `betas=(0.9,0.999)` and `epsilon=1e-08`
|
| 46 |
+
- `lr_scheduler_type`: linear
|
| 47 |
+
- `lr_scheduler_warmup_ratio`: 0.1
|
| 48 |
+
- `num_epochs`: 10.0
|
| 49 |
+
- `mixed_precision_training`: Native AMP
|
| 50 |
|
| 51 |
### Training results
|
| 52 |
|
| 53 |
+
| Training Loss | Epoch | Step | Validation Loss | F1 | Roc Auc | Accuracy | Map |
|
| 54 |
+
| :-----------: | :---: | :---: | :-------------: | :----: | :-----: | :------: | :----: |
|
| 55 |
+
| 1.5521 | 1.0 | 153 | 0.7759 | 0.3929 | 0.6789 | 0.0209 | 0.3394 |
|
| 56 |
+
| 0.7088 | 2.0 | 306 | 0.5183 | 0.4480 | 0.7162 | 0.0349 | 0.4047 |
|
| 57 |
+
| 0.484 | 3.0 | 459 | 0.4342 | 0.4673 | 0.7241 | 0.0447 | 0.4348 |
|
| 58 |
+
| 0.369 | 4.0 | 612 | 0.3847 | 0.4777 | 0.7332 | 0.0504 | 0.4463 |
|
| 59 |
+
| 0.2943 | 5.0 | 765 | 0.3587 | 0.4838 | 0.7284 | 0.0572 | 0.4556 |
|
| 60 |
+
| 0.2446 | 6.0 | 918 | 0.3415 | 0.4875 | 0.7296 | 0.0608 | 0.4628 |
|
| 61 |
+
| 0.2099 | 7.0 | 1071 | 0.3273 | 0.4896 | 0.7246 | 0.0648 | 0.4682 |
|
| 62 |
+
| 0.186 | 8.0 | 1224 | 0.3140 | 0.4888 | 0.7171 | 0.0689 | 0.4711 |
|
| 63 |
+
| 0.1693 | 9.0 | 1377 | 0.3101 | 0.4887 | 0.7157 | 0.0703 | 0.4741 |
|
| 64 |
+
| 0.1582 | 10.0 | 1530 | 0.3063 | 0.4876 | 0.7140 | 0.0714 | 0.4743 |
|
| 65 |
+
|
| 66 |
+
## Disclaimer
|
| 67 |
+
|
| 68 |
+
Do consider the biases which came from pre-training datasets that may be carried over into the results of this model.
|
| 69 |
+
|
| 70 |
+
## Authors
|
| 71 |
+
|
| 72 |
+
Distil Audio Spectrogram Transformer AudioSet was trained and evaluated by [Ananto Joyoadikusumo](https://anantoj.github.io), [David Samuel Setiawan](https://davidsamuell.github.io/), [Wilson Wongso](https://wilsonwongso.dev/). All computation and development are done on Google Cloud.
|
| 73 |
+
|
| 74 |
+
## Framework versions
|
| 75 |
|
| 76 |
- Transformers 4.27.0.dev0
|
| 77 |
- Pytorch 1.13.1+cu117
|