tools/tts_sample.py

"""
tts_sample.py — genera y reproduce una muestra de las 3 voces (lesky, monstruo,
gnomos) llamando al endpoint Modal. Guarda los WAV en samples/ y los reproduce
en orden (macOS `afplay`). Sirve para oír cómo quedan las clonaciones.

Uso:
  python tools/tts_sample.py
  python tools/tts_sample.py "https://otra-url.modal.run"   # URL distinta (opcional)

OJO: si una voz suena neutra (no clonada), falta `modal deploy tools/tts_modal.py`
tras añadir su .mp3 — los conds se cargan al arrancar el contenedor.
"""

import os
import sys
import time
import json
import subprocess
import urllib.request

URL = sys.argv[1] if len(sys.argv) > 1 else \
    "https://vegagallegopablo--chatterbox-tts-turbo-tts-speak.modal.run"

SAMPLES = {
    "lesky":    "Heh heh heh! Welcome to my shop, little one. The treasure? Oh, it's mine.",
    "monstruo": "So... another little challenger crawls into my alley. How adorable. I will crush you.",
    "gnomos":   "Wait, wait, no! It was MY turn to be on top! Move over, you're squashing my hat!",
}

OUT_DIR = "samples"
os.makedirs(OUT_DIR, exist_ok=True)


def synth(text, voice):
    d = json.dumps({"text": text, "voice": voice}).encode()
    req = urllib.request.Request(URL, data=d, headers={"Content-Type": "application/json"})
    t = time.time()
    with urllib.request.urlopen(req, timeout=180) as r:
        audio = r.read()
    return audio, time.time() - t


for voice, text in SAMPLES.items():
    print(f"\n🎙  {voice}: «{text}»")
    try:
        audio, dt = synth(text, voice)
    except Exception as e:
        print(f"   ❌ fallo: {e}")
        continue
    path = os.path.join(OUT_DIR, f"sample_{voice}.wav")
    with open(path, "wb") as f:
        f.write(audio)
    print(f"   ✅ {len(audio)} bytes · {dt:.2f}s · {path}")
    try:
        subprocess.run(["afplay", path])    # suena por los altavoces (macOS)
    except FileNotFoundError:
        print("   (sin afplay: reprodúcelo a mano)")

print(f"\nListo. WAV en {OUT_DIR}/")