metadata

datasets:
  - facebook/multilingual_librispeech
language:
  - it
base_model:
  - SWivid/F5-TTS
pipeline_tag: text-to-speech
license: cc-by-4.0
library_name: f5-tts

This is an Italian finetune for F5-TTS

UPDATE:

A better version with improved prosody here => https://huggingface.co/alien79/F5-TTS-italian *

Italian only so can't speak english properly

Trained over 247+h hours of "train" split of facebook/multilingual_librispeech dataset, 6717 steps for Epoch:

catastrophic failure (the model forgot english)
italian pronunciation not perfect (there are lot of checkpoints to let you play with and extend training, maybe with different datasets)

Current most trained model

italian_59kh/model_464400.safetensors (~70 Epoch)

folder structure:

| - italian_59kh
|   | - checkpoints

italian_59kh

Contains the weight at specific steps, the higher the number, the further it went into training. Weights in this folder can't be used to resume training, use checkpoints instead.

italian_59kh/checkpoints

Contains the weight of the checkpoints at specific steps, the higher the number, the further it went into training. Weights in this folder can be used as starting point to continue training.

The run.py file is an example of how to extract the wav files and produce the metadata.csv to use for training