app.py

import gradio as gr
from TTS.api import TTS
from fastapi import FastAPI, Request
from fastapi.responses import FileResponse, JSONResponse
import uvicorn
import time
import os

# -----------------------------
# Configuration
# -----------------------------
YOURTTS_MODEL = "tts_models/multilingual/multi-dataset/your_tts"
SPEAKERS = {
    "male": "speakers/voice1.wav",
    "female": "speakers/voice2.wav"
}
OUTPUT_PATH = "output.wav"

tts = TTS(YOURTTS_MODEL, gpu=False)

# -----------------------------
# Core synthesis function
# -----------------------------
def synthesize(text: str, speaker: str = "female"):
    speaker_path = SPEAKERS.get(speaker.lower())
    if not speaker_path or not os.path.exists(speaker_path):
        return None, {"error": f"❌ Speaker file not found: {speaker_path}"}

    start_time = time.time()
    try:
        tts.tts_to_file(
            text=text,
            speaker_wav=speaker_path,
            file_path=OUTPUT_PATH,
            language="en"
        )
    except Exception as e:
        return None, {"error": str(e)}

    total_time = time.time() - start_time
    est_duration = len(text.split()) / 2.5
    rtf = round(total_time / est_duration, 3)
    info = {
        "language": "English",
        "processing_time_sec": round(total_time, 3),
        "real_time_factor": rtf,
        "model_used": YOURTTS_MODEL,
        "speaker_used": os.path.basename(speaker_path),
    }
    return OUTPUT_PATH, info

# -----------------------------
# FastAPI setup
# -----------------------------
app = FastAPI(title="YourTTS FastAPI", description="Text-to-Speech API")

@app.post("/synthesize")
async def predict(request: Request):
    data = await request.json()
    text = data.get("text")
    speaker = data.get("speaker", "female")
    if not text:
        return JSONResponse({"error": "Missing 'text' field"}, status_code=400)

    audio_path, info = synthesize(text, speaker)
    if audio_path is None:
        return JSONResponse(info, status_code=500)

    headers = {f"x-{k}": str(v) for k, v in info.items()}
    return FileResponse(audio_path, media_type="audio/wav", filename="output.wav", headers=headers)

# -----------------------------
# Gradio UI
# -----------------------------
demo = gr.Interface(
    fn=synthesize,
    inputs=[
        gr.Textbox(label="Text"),
        gr.Dropdown(choices=["male", "female"], value="female", label="Speaker")
    ],
    outputs=[gr.Audio(type="filepath"), gr.JSON()],
    title="YourTTS Voice Cloning (English Only, Select Speaker)",
    allow_flagging="never"
)

# -----------------------------
# Run both FastAPI + Gradio
# -----------------------------
if __name__ == "__main__":
    import threading

    def launch_gradio():
        demo.launch(server_name="0.0.0.0", server_port=7861, show_api=False)

    threading.Thread(target=launch_gradio, daemon=True).start()

    uvicorn.run(app, host="0.0.0.0", port=7860)