Create voice samples from a written description

import base64
import json
import os
from pathlib import Path
import requests

response = requests.post("https://api.nativeport.ai/fishaudio/v1/voice-design",
    headers={"Authorization": f"Bearer {os.environ['NATIVEPORT_API_KEY']}",
             "model": "voice-design-1"},
    json={"instruction": "A warm, calm English-speaking narrator with clear diction and a relaxed pace.",
          "reference_text": "Welcome. Take a moment to settle in, and we will guide you through the next steps.",
          "language": "en", "n": 2}, timeout=120)
response.raise_for_status()
candidates = response.json().get("candidates") or []
if not candidates:
    raise RuntimeError("No voice samples returned")
metadata = []
for index, candidate in enumerate(candidates, start=1):
    audio = base64.b64decode(candidate["audio_base64"], validate=True)
    if not audio:
        raise RuntimeError("Empty candidate audio")
    if audio.startswith(b"RIFF") and audio[8:12] == b"WAVE":
        extension = "wav"
    elif audio.startswith(b"OggS"):
        extension = "ogg"
    elif audio.startswith(b"ID3") or (len(audio) > 1 and audio[0] == 255 and audio[1] & 224 == 224):
        extension = "mp3"
    else:
        extension = "bin"  # Inspect the encoding before choosing a player.
    filename = f"voice-sample-{index}.{extension}"
    Path(filename).write_bytes(audio)
    metadata.append({"file": filename, **{key: value for key, value in candidate.items()
                                          if key != "audio_base64"}})
Path("voice-samples.json").write_text(json.dumps(metadata, indent=2), encoding="utf-8")
print(json.dumps(metadata, indent=2))