奈亚
目录
  1. 1视频手册
  2. 2Naia OS Live USB
  3. 3安装部署
  4. 3.1Naia OS 安装 (ISO)
  5. 3.2应用安装
  6. 4开始使用
  7. 4.1Naia Model Pricing
  8. 4.2naia-omni-cascade
  9. 4.3demo
  10. 4.4naia-offline
  11. 4.5naia-model-dev
  12. 4.6naia-online
  13. 5主屏幕
  14. 6聊天
  15. 6.1个人电台 DJ 与展览导览
  16. 7对话历史记录
  17. 8工作进展
  18. 9技能
  19. 10渠道
  20. 11代理商
  21. 12诊断
  22. 13工作区
  23. 14浏览器
  24. 15面板管理
  25. 16语音对话
  26. 17设置
  27. 18工具详情
  28. 19Naia 账号
  29. 20故障排除
  30. 21开源使用与贡献

6.1. 个人电台 DJ 与展览导览

Naia has two opt-in modes that can speak before you type: Personal Radio DJ and Exhibition Guide.

Start and stop

Enter one of these exact phrases in chat.

ModeStartStop
Personal Radio DJ개인 라디오 시작해 or radio DJ start라디오 종료
Exhibition Guide행사 소개 시작 or exhibition start행사 소개 종료

The selection is saved for the current workspace. New user input takes priority over proactive speech.

Personal Radio DJ

After a default 5-second idle delay, Naia starts YouTube BGM and gives short DJ remarks. The normal settings screen does not currently expose this delay.

Korean voice controls currently include 음악만 (music only), 말 줄여 (talk less), 계속 이야기해 (talk more), 분위기 바꿔 (change vibe), 다음 곡 (next), and 그만해 (stop).

The current version uses time, current BGM, optional injected weather, and explicit preferences held in the running process. Mood/activity collection, user-facing weather consent and location settings, and durable Naia Memory music preferences are not connected yet. Remarks rotate through three safe phrases; this is not yet free-form, long-running radio-quality hosting.

Exhibition Guide

After a default 1-second idle delay, Naia greets visitors and introduces items from the configured company, product, and event knowledge source. The normal settings screen does not currently expose this delay. It abstains when that source does not support an answer.

Korean controls include 조용히 (quiet), 소개 계속 (resume), 처음부터 소개 (restart), and 그만해 (stop). Visitor questions are designed to interrupt the introduction and resume at the next item. Exhibition activity and Q&A do not use long-term memory or persistent transcripts by default.

Current validation boundary

Native Tauri tests currently prove profile persistence/restoration, real YouTube BGM plus the first DJ result and stop, and the exhibition greeting and stop. Audible proactive TTS, two live DJ remarks, live question interruption/resume, every control, long-duration behavior, and field quality are not yet product-validated.

Different from AI interjections: AI interjections allow suggestions during a conversation. These two profiles are separate long-running activities that you explicitly enable with a start phrase.