metadata

library_name: transformers
language:
  - twi
license: mit
base_model: microsoft/speecht5_tts
tags:
  - generated_from_trainer
datasets:
  - fsicoli/twi
model-index:
  - name: SpeechT5 TTS Npontu Twi
    results: []

SpeechT5 TTS Npontu Twi

This model is a fine-tuned version of microsoft/speecht5_tts on the FsicoliTwi dataset. It achieves the following results on the evaluation set:

Loss: 0.3824

Model description

Npontu Twi is designed to synthesize Twi-language speech with a focus on Ghanaian accents and cultural nuances. Leveraging pure language modeling, Npontu Twi offers high-quality, natural, and culturally relevant speech synthesis for diverse applications, including education, entertainment, and communication in Ghana and beyond.

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: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 2000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.4207	14.4928	1000	0.3869
0.41	28.9855	2000	0.3824

Framework versions

Transformers 4.49.0.dev0
Pytorch 2.5.1+cu121
Datasets 3.2.0
Tokenizers 0.21.0