Clonare LA TUA VOCE OFFLINE con l'IA (Tutorial Piper TTS)

Описание: Ciao, in questo video ti mostro come addestrare un modello TTS (Text-to-Speech) incredibilmente realistico usando Piper TTS, girando tutto al 100% in locale sul tuo PC.


LINK E RISORSE:

GUIDA COMPLETA: https://www.francescogarofalo.it/post...

Repository Ufficiale Piper: https://github.com/OHF-voice/piper1-g...

Dataset usato: https://huggingface.co/datasets/rhass...

CONTATTI:
Blog: https://www.francescogarofalo.it/ping...
Instagram:   / francesco182g  

I comandi chiave usati nel video:

Clona e installa Piper:
git clone https://github.com/OHF-voice/piper1-g...
cd piper1-gpl
python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install -e .[train]



Eseguire il Comando di Training

python3 -m piper.train fit \
--data.voice_name "my_italian_voice" \
--data.csv_path /home/user/my-voice-dataset/metadata.csv \
--data.audio_dir /home/user/my-voice-dataset/wavs/ \
--model.sample_rate 22050 \
--data.espeak_voice "it" \
--data.cache_dir /home/user/my-voice-dataset/cache/ \
--data.config_path /home/user/my-voice-dataset/config.json \
--data.batch_size 32 \
--ckpt_path /home/user/piper-checkpoints/epoch=2000.ckpt

Monitorare l’avanzamento:

tensorboard --logdir tensorboard --logdir /mnt/d/Projects/piper/piper1-gpl/lightning_logs


Esportare la Tua Voce

python3 -m piper.train.export_onnx \
--checkpoint /home/user/my-voice-dataset/logs/lightning_logs/version_0/checkpoints/last.ckpt \
--output-file /home/user/my_italian_voice.onnx

Grazie :)