feat(tts): add aivis_tts_python (Aivis Cloud, Japanese realtime HTTP streaming)

[Therealratoshen]

Summary

Adds a new TTS extension aivis_tts_python backed by Aivis Cloud API and wires it into the voice-assistant example as voice_assistant_aivis.

Aivis is a Japanese-focused realtime TTS provider with public HTTP API and WAV streaming — fills the regional TTS gap (Supertone sunsets 2026-08-31, signup already closed).

Changes

• ai_agents/agents/ten_packages/extension/aivis_tts_python/ — new extension modeled on the existing Gradium/Rime HTTP TTS extensions:
  • AsyncTTS2HttpExtension subclass streaming POST /v1/tts/synthesize via httpx.AsyncClient.
  • WavStreamParser strips the WAV header so the first PCM chunk reaches TEN immediately, giving the base class a correct TTFB measurement.
  • 401/403 → INVALID_KEY_ERROR; cancellation → FLUSH; other HTTP errors → ERROR.
  • output_format is forced to wav because TEN needs raw PCM.
• ai_agents/agents/examples/voice-assistant/tenapp/property.json — appends voice_assistant_aivis (Deepgram ja STT + OpenAI LLM + Aivis TTS + main_control + message_collector).
• ai_agents/.env.example — documents AIVIS_API_KEY, AIVIS_MODEL_UUID, AIVIS_BASE_URL alongside the other TTS providers.

Test plan

• tests/test_config.py — 4 unit tests (validation, default param normalization, client-only key stripping, URL construction) pass without TEN runtime.
• Run tests/bin/start end-to-end in a TEN checkout where the runtime has been built (task install && task build in ai_agents/) — this requires Docker on macOS and so could not be done in this PR's environment. Same prerequisite as every other TTS extension's tests (verified against Gradium).

Notes

• Default AIVIS_MODEL_UUID points at Aivis's public default model (a59cb814-0083-4369-8542-f51a29e72af7).
• No secrets in tree; all keys come from environment.
• License: Apache-2.0 (matches the rest of the repo).

Made with Cursor