metadata
license: mit
base_model: microsoft/speecht5_tts
tags:
- generated_from_trainer
- text-to-speech
datasets:
- facebook/voxpopuli
model-index:
- name: speecht5_finetuned_voxpopuli_fr
results: []
speecht5_finetuned_voxpopuli_fr
This model is a fine-tuned version of microsoft/speecht5_tts on the French part of facebook/voxpopuli dataset. It achieves the following results on the evaluation set:
- Loss: 0.4644
It was trained on only a small fraction of the original dataset !
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 4
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.5567 | 1.49 | 500 | 0.5125 |
| 0.5272 | 2.99 | 1000 | 0.4876 |
| 0.516 | 4.48 | 1500 | 0.4767 |
| 0.5066 | 5.98 | 2000 | 0.4709 |
| 0.5072 | 7.47 | 2500 | 0.4669 |
| 0.4994 | 8.97 | 3000 | 0.4644 |
| 0.4964 | 10.46 | 3500 | 0.4652 |
| 0.4999 | 11.96 | 4000 | 0.4644 |
Framework versions
- Transformers 4.38.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0