app.py

import gradio as gr
from transformers import pipeline

# Define a list of available speaker models
SPEAKER_MODELS = {
    "Default": "onnx-community/Kokoro-82M-ONNX",
    "Speaker 1": "onnx-community/Kokoro-82M-ONNX",
    "Speaker 2": "onnx-community/Kokoro-82M-ONNX"
}

def kokoro_tts(text, speaker):
    # Initialize the transformers pipeline for text-to-speech with the selected speaker model
    tts_pipeline = pipeline("text-to-speech", model=SPEAKER_MODELS[speaker])
    # Generate speech from text
    speech = tts_pipeline(text)
    return speech["audio"]

# Create a Gradio interface
iface = gr.Interface(
    fn=kokoro_tts,
    inputs=[
        gr.Textbox(lines=2, placeholder="Enter text here..."),
        gr.Dropdown(choices=list(SPEAKER_MODELS.keys()), label="Select Speaker")
    ],
    outputs=gr.Audio(label="Generated Speech"),
    title="Kokoro Text-to-Speech",
    description="A Text-to-Speech app powered by Hugging Face Transformers.js with multiple speaker options"
)

if __name__ == "__main__":
    iface.launch()