MMS TTS THAI MALEV1

🚀 Thai Text-to-Speech Model

This project is a Thai text-to-speech model that uses audio recordings from Play.ht to finetune the model, enabling high - quality Thai speech synthesis.

🚀 Quick Start

This model uses audio recordings from Play.ht: https://play.ht/ to finetune the model.

📦 Installation

• Finetune the model: You can find the GitHub code for finetuning the model at: https://github.com/VYNCX/finetune-local-vits
• Train the model on Google Colab: Use Finetune - colab to train the speech model on Google Colab by yourself.
• Run locally: Run the model on your local computer with the code at: https://github.com/VYNCX/VachanaTTS

💻 Usage Examples

Basic Usage

import torch
from transformers import VitsTokenizer, VitsModel, set_seed
import scipy

tokenizer = VitsTokenizer.from_pretrained("VIZINTZOR/MMS-TTS-THAI-MALEV1",cache_dir="./mms")
model = VitsModel.from_pretrained("VIZINTZOR/MMS-TTS-THAI-MALEV1",cache_dir="./mms")

inputs = tokenizer(text="สวัสดีครับ นี่คือเสียงพูดภาษาไทย", return_tensors="pt")

set_seed(456)  # make deterministic

with torch.no_grad():
   outputs = model(**inputs)

waveform = outputs.waveform[0]

# Convert PyTorch tensor to NumPy array
waveform_array = waveform.numpy()

scipy.io.wavfile.write("techno_output.wav", rate=model.config.sampling_rate, data=waveform_array)