app.py

import numpy as np
from fastapi import FastAPI
from fastapi.responses import StreamingResponse
import io
from gtts import gTTS
import tempfile
import wave

app = FastAPI()

def generate_speech_pcm():
    text = "Hello, this is ESP32 audio test"

    tts = gTTS(text=text, lang='en')

    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as f:
        tts.save(f.name)

        # Convert MP3 to PCM using wave fallback
        import subprocess
        wav_file = f.name.replace(".mp3", ".wav")

        subprocess.run([
            "ffmpeg", "-y",
            "-i", f.name,
            "-ar", "24000",
            "-ac", "1",
            "-f", "wav",
            wav_file
        ], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)

        with wave.open(wav_file, "rb") as w:
            frames = w.readframes(w.getnframes())

    return frames

@app.get("/audio")
def stream_audio():
    return StreamingResponse(
        io.BytesIO(generate_speech_pcm()),
        media_type="application/octet-stream"
    )

import uvicorn

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=7860)