app.py

import spaces
import gradio as gr
from kokoro import KPipeline
import soundfile as sf
import tempfile

# Set up the Kokoro pipeline (choose your language code)
pipeline = KPipeline(lang_code='a')  # 'a' = American English, see docs for more

@spaces.GPU
def kokoro_tts(text: str, voice: str = "af_heart", speed: float = 1.0) -> str:
    """
    Generate speech audio from text using Kokoro TTS.
    Args:
        text: The input text to synthesize.
        voice: The Kokoro voice name (e.g., 'af_heart').
        speed: Speech speed (default 1.0).
    Returns:
        Path to the generated WAV audio file.
    """
    # Generate audio using Kokoro
    generator = pipeline(text, voice=voice, speed=speed)
    for i, (_, _, audio) in enumerate(generator):
        # Save the first audio chunk as a WAV file
        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
            sf.write(f.name, audio, 24000)
            return f.name  # Return file path for Gradio to serve

# List some example voices
voice_options = ["af_heart", "af_bella", "af_aria", "af_riley", "af_ryan"]

description = """
Enter text to synthesize speech using Kokoro TTS.<br>
**Voices:** af_heart, af_bella, af_aria, af_riley, af_ryan<br>
**Languages:** American English (default), see Kokoro docs for more.<br>
**Speed:** 1.0 = normal, lower = slower, higher = faster.<br>
"""

demo = gr.Interface(
    fn=kokoro_tts,
    inputs=[
        gr.Textbox(label="Text"),
        gr.Dropdown(voice_options, label="Voice", value="af_heart"),
        gr.Slider(0.5, 2.0, value=1.0, label="Speed")
    ],
    outputs=gr.Audio(type="filepath", label="Generated Speech"),
    title="Kokoro TTS MCP Server",
    description=description
)

if __name__ == "__main__":
    demo.launch(mcp_server=True)