<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
STT_Model_9
This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2506
- Wer: 0.1718
Model description
More information needed
Intended uses & limitations
More information needed
Dataset info
- Name: LJSpeech
- Source: https://www.kaggle.com/datasets/mathurinache/the-lj-speech-dataset
- Total audios (in Google Drive): 1420
- Total transcripts (in Google Drive): 13100
- No. of rows selected: 500
- Train-test ratio: 70:30
- No. of training set: 350
- No. of testing set: 150
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| No log | 4.55 | 200 | 2.9217 | 0.9846 |
| No log | 9.09 | 400 | 1.2293 | 0.7093 |
| 2.3111 | 13.64 | 600 | 0.3885 | 0.3602 |
| 2.3111 | 18.18 | 800 | 0.3123 | 0.3097 |
| 0.2471 | 22.73 | 1000 | 0.3094 | 0.2737 |
| 0.2471 | 27.27 | 1200 | 0.3007 | 0.2537 |
| 0.2471 | 31.82 | 1400 | 0.2650 | 0.2008 |
| 0.0853 | 36.36 | 1600 | 0.2599 | 0.1884 |
| 0.0853 | 40.91 | 1800 | 0.2462 | 0.1734 |
| 0.0344 | 45.45 | 2000 | 0.2663 | 0.1730 |
| 0.0344 | 50.0 | 2200 | 0.2506 | 0.1718 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.13.1+cu116
- Datasets 2.9.0
- Tokenizers 0.13.2
