| import tempfile ,os | |
| import gradio as gr | |
| from transformers import VitsModel, AutoTokenizer,pipeline | |
| import torch | |
| import numpy as np | |
| import torchaudio | |
| model = VitsModel.from_pretrained("SeyedAli/Persian-Speech-synthesis") | |
| tokenizer = AutoTokenizer.from_pretrained("SeyedAli/Persian-Speech-synthesis") | |
| text_input = gr.TextArea(label="متن فارسی",text_align="right",rtl=True,type="text") | |
| audio_output = gr.Audio(label="صوت گفتار فارسی", type="filepath") | |
| def TTS(text): | |
| inputs = tokenizer(text, return_tensors="pt") | |
| with torch.no_grad(): | |
| output = model(**inputs).waveform | |
| with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp: | |
| torchaudio.save(fp, output, model.config.sampling_rate,format="wav") | |
| return fp.name | |
| iface = gr.Interface(fn=TTS, inputs=text_input, outputs=audio_output) | |
| iface.launch(share=False) |