license: mit
base_model: microsoft/speecht5_tts
tags:
- generated_from_trainer
- text-to-speech
datasets:
- voxpopuli
model-index:
- name: speecht5_finetuned_voxpopuli_nl
results: []
speecht5_finetuned_voxpopuli_nl
This model is a fine-tuned version of microsoft/speecht5_tts on the italian section of voxpopuli dataset. It achieves the following results on the evaluation set:
- Loss: 0.4896
Model description
This model do text to speeche task in italian language
Intended uses & limitations
More information needed Example:
from transformers import AutoProcessor, SpeechT5ForTextToSpeech
processor = AutoProcessor.from_pretrained("jjsprockel/speecht5_finetuned_voxpopuli_it") model = SpeechT5ForTextToSpeech.from_pretrained("jjsprockel/speecht5_finetuned_voxpopuli_it")
speaker_embeddings = torch.tensor(example["speaker_embeddings"]).unsqueeze(0)
text = "Quando pensi che sarà possibile viaggiare?" inputs = processor(text=text, return_tensors="pt")
vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan") speech = model.generate_speech(inputs["input_ids"], speaker_embeddings, vocoder=vocoder)
from IPython.display import Audio Audio(speech.numpy(), rate=16000)
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 |
|---|---|---|---|
| 0.5445 | 6.13 | 1000 | 0.5106 |
| 0.5262 | 12.26 | 2000 | 0.4964 |
| 0.5154 | 18.39 | 3000 | 0.4918 |
| 0.5186 | 24.52 | 4000 | 0.4896 |
Framework versions
- Transformers 4.31.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3