---
license: mit
base_model: microsoft/speecht5_tts
tags:
- text-to-speech
datasets:
- facebook/voxpopuli
model-index:
- name: speecht5_tts-ft-voxpopuli-it
  results:
  - task:
      type: text-to-speech
    dataset:
      name: facebook/voxpopuli
      type: facebook/voxpopuli
      config: it
      split: train
      args: it
    metrics:
    - name: N.A.
      type: N.A.
      value: N.A.
language:
- it
---



# speecht5_tts-ft-voxpopuli-it

This model is a fine-tuned version of [microsoft/speecht5_tts](https://huggingface.co/microsoft/speecht5_tts) on the facebook/voxpopuli dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5126

## Model description

It uses the speaker embedding model speechbrain/spkrec-xvect-voxceleb

## Intended uses & limitations

More information needed

## Training and evaluation data

test_size=0.15

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 8
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 4
- 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: 300
- training_steps: 1000

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 0.6118        | 1.94  | 300  | 0.5508          |
| 0.5729        | 3.89  | 600  | 0.5204          |
| 0.563         | 5.83  | 900  | 0.5126          |


### Framework versions

- Transformers 4.33.0
- Pytorch 1.12.1+cu116
- Datasets 2.14.4
- Tokenizers 0.12.1