app.py

# This Gradio app generates audio from text using a simple text-to-speech model with configurable generation parameters.
import gradio as gr
import numpy as np

# Define a function that takes text and generates audio with configurable parameters.
def text_to_audio(text, use_sampling=True, top_k=250, top_p=0.0, temperature=1.0, duration=10.0, cfg_coef=3.0):
    # For demonstration purposes, we'll generate a simple sine wave.
    # In a real application, you would use a text-to-speech model here.
    sr = 44100  # Sample rate
    frequency = 440  # Frequency in Hz (A4 note)
    t = np.linspace(0, duration, sr * duration, endpoint=False)
    audio = (0.5 * np.sin(2 * np.pi * frequency * t)).astype(np.float32)
    return (sr, audio)

# Create a Gradio interface that takes a textbox input, runs it through the text_to_audio function, and returns output to an audio component.
with gr.Blocks() as demo:
    with gr.Row():
        text_input = gr.Textbox(label="Input Text")
        use_sampling_checkbox = gr.Checkbox(label="Use Sampling", value=True)
    with gr.Row():
        top_k_slider = gr.Slider(1, 1000, value=250, label="Top K")
        top_p_slider = gr.Slider(0.0, 1.0, value=0.0, label="Top P")
    with gr.Row():
        temperature_slider = gr.Slider(0.1, 10.0, value=1.0, label="Temperature")
        duration_slider = gr.Slider(1.0, 60.0, value=10.0, label="Duration (s)")
        cfg_coef_slider = gr.Slider(0.1, 10.0, value=3.0, label="CFG Coefficient")
    audio_output = gr.Audio(label="Generated Audio", type="numpy")
    
    text_input.change(
        fn=text_to_audio,
        inputs=[text_input, use_sampling_checkbox, top_k_slider, top_p_slider, temperature_slider, duration_slider, cfg_coef_slider],
        outputs=audio_output
    )

# Launch the interface.
demo.launch(show_error=True)