22 January 2025
I use Piper as TTS for my project, but at the time there wasn’t a public checkpoint for the Italian language so I decided to create my own.
Voices are trained using the Multilingual LibriSpeech (MLS)
The resulting checkpoints and onnx models are available in this huggingface repository
Voices
-
Aurora: From scratch
-
Sample 1:
-
Sample 2:
-
Sample 3:
-
-
Leonardo: From scratch
-
Sample 1:
-
Sample 2:
-
Sample 3:
-
-
Giorgio: Fine tune of Leonardo
-
Sample 1:
-
Sample 2:
-
Sample 3:
-