Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html>
<h"... is not valid JSON
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Cores — Français (fr)
LJSpeech dataset of Cores (fr).
- 327 pairs
- 44.1kHz mono 16-bit PCM WAV (original wiki quality)
Piper TTS Training (High Quality on T4 GPU)
Preprocessing (downsample to 22.05kHz)
python3 -m piper_train.preprocess \
--language fr \
--input-dir ./cores_fr \
--output-dir ./train_cores_fr \
--dataset-format ljspeech \
--single-speaker \
--sample-rate 22050
Training (Kaggle T4 16GB)
python3 -m piper_train \
--dataset-dir ./train_cores_fr \
--accelerator 'gpu' \
--devices 1 \
--quality high \
--batch-size 16 \
--validation-split 0.0 \
--num-test-examples 0 \
--max_epochs 10000 \
--resume_from_checkpoint /path/to/checkpoint.ckpt \
--checkpoint-epochs 1 \
--precision 32 \
--max-phoneme-ids 400
- Downloads last month
- 32