app.py

from numpy import int16,float32
import gradio as gr
from transformers import AutoModel
model = AutoModel.from_pretrained("ai4bharat/IndicF5", trust_remote_code=True).to("cpu")
def synthesize_speech(text, ref_audio, ref_text):
    audio = model(text, ref_audio_path=ref_audio, ref_text=ref_text)
    if audio.dtype == int16:
        audio = audio.astype(float32) / 32768.0
    return 24000, audio
gr.Interface(synthesize_speech,[gr.Textbox(label="Text to Synthesize"), gr.Audio(type="filepath", label="Reference Prompt Audio"), gr.Textbox(label="Text in Reference Prompt Audio")], gr.Audio(label="Generated Speech", type="numpy")).launch()