"""Text-to-speech serving app (VoxCPM2 — a 2B MiniCPM-4-backbone TTS). Deploy: modal deploy modal/app_tts.py (or: uv run scripts/deploy_modal.py tts) Serve dev: modal serve modal/app_tts.py Each model in the media catalogue gets its own OpenAI-compatible ``/v1/audio/speech`` endpoint at ``…--audio-tts-.modal.run``. """ from __future__ import annotations import modal from media_catalogue import TTS_APP, TTS_MODELS from tts_service import register_all app = modal.App(TTS_APP) register_all(app, TTS_MODELS)