services/tts_service.py

from google.genai import types
from services.gemini_client import get_gemini_client
import wave


def save_wave_file(filename, pcm, channels=1, rate=24000, sample_width=2):
    with wave.open(filename, "wb") as wf:
        wf.setnchannels(channels)
        wf.setsampwidth(sample_width)
        wf.setframerate(rate)
        wf.writeframes(pcm)


async def generate_tts(text: str) -> bytes:
    client = get_gemini_client()

    response = client.models.generate_content(
        model="gemini-2.5-flash-preview-tts",
        contents=text,
        config=types.GenerateContentConfig(
            response_modalities=["AUDIO"],
            speech_config=types.SpeechConfig(
                voice_config=types.VoiceConfig(
                    prebuilt_voice_config=types.PrebuiltVoiceConfig(
                        voice_name="Kore"
                    )
                )
            ),
        ),
    )

    audio_bytes = response.candidates[0].content.parts[0].inline_data.data
    return audio_bytes