BIA-XTTSv2-moore-v0

Model Description

BIA-XTTSv2-moore-v0 is a fine-tuned version of Coqui's XTTS v2 model specifically trained for Moore language, a Gur language spoken in Burkina Faso. This model enables high-quality text-to-speech synthesis for Moore speakers and supports various voice cloning capabilities.

Language Information

• Language: Moore (Mooré)
• Language Code: mos
• Region: Burkina Faso, West Africa
• Speakers: ~5 million native speakers
• Language Family: Niger-Congo → Gur → Oti-Volta → Moore

Usage

import torch, torchaudio, os
import numpy as np

from   tqdm import tqdm
from TTS.tts.configs.xtts_config import XttsConfig
from TTS.tts.models.xtts         import Xtts

checkpoint_path = "checkpints"
model_path      = "best_model.pth"

device = "cuda:0" if torch.cuda.is_available() else "cpu"

xtts_checkpoint = os.path.join(checkpoint_path, model_path)
xtts_config     = os.path.join(checkpoint_path,"config.json")
xtts_vocab      = checkpoint_path+"vocab.json"

# Load model
config     = XttsConfig()
config.load_json(xtts_config)
XTTS_MODEL = Xtts.init_from_config(config)
XTTS_MODEL.load_checkpoint(config,
                           checkpoint_path = xtts_checkpoint,
                           vocab_path      = xtts_vocab,
                          

Model Details

Training Data

• Dataset: Curated Moore speech dataset
• Hours: High-quality Moore audio recordings 50
• Speakers: Multiple native Moore speakers
• Quality: Studio and field recordings, carefully preprocessed

Training Configuration

• Base Model: XTTS v2
• Fine-tuning Steps: 9,000+ steps OF BACT
• Batch Size: 2 and gradient accumulation of 4
• Learning Rate: Adaptive scheduling
• Hardware: GPU-accelerated training

Citation

If you use this model in your research, please cite:

@misc{bia-xtts-moore-v0,
  title={BIA-XTTSv2-moore-v0: A Fine-tuned XTTS Model for Moore Language},
  author={BIA Team},
  year={2024},
  howpublished={\url{https://huggingface.co/BIA/BIA-XTTSv2-moore-v0}}
}

Acknowledgments

• Coqui TTS team for the original XTTS v2 architecture
• Moore language community in Burkina Faso
• Data contributors who provided high-quality recordings