Speecht5 Tts KTH Hu

🚀 Experimental SpeechT5 Fine - Tuning for Hungarian

This project is an experimental fine - tuning of SpeechT5 for the Hungarian language. It aims to provide high - quality text - to - speech conversion for Hungarian users.

🚀 Quick Start

Prerequisites

Make sure you have installed the necessary libraries such as transformers, datasets, torch, and IPython.

Usage Example

from transformers import SpeechT5Processor, SpeechT5ForTextToSpeech, SpeechT5HifiGan
from datasets import load_dataset
import torch
from IPython.display import Audio

processor = SpeechT5Processor.from_pretrained("microsoft/speecht5_tts")
model = SpeechT5ForTextToSpeech.from_pretrained("GaborMadarasz/speecht5_tts_KTH_hu")
vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan")

inputs = processor(text="Azóta, hogy nem láttuk őket, nyolc esztendő telt el.", return_tensors="pt")

# load xvector containing speaker's voice characteristics from a dataset
embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
speaker_embeddings = torch.tensor(embeddings_dataset[7406]["xvector"]).unsqueeze(0)

speech = model.generate_speech(inputs["input_ids"], speaker_embeddings, vocoder=vocoder)

# Jupyter audio 
Audio(speech, rate=16000)

📄 License

This project is licensed under the MIT license.

📦 Information Table