A Karaoke Server for My Partner, and Two Local AI Models That Argue About Cantopop at 4 am

Описание и применение

Создан автоматизированный сервер караоке, который позволяет пользователям добавлять песни через QR‑код, отображает тексты и автоматически классифицирует новые файлы по языку с помощью локальных моделей ИИ.

Задача

The partner’s karaoke machine had a slow, locked‑down tablet that made searching for songs in Mandarin, Cantonese, and Korean difficult, and the existing song library was poorly organized with untidy filenames.

Результат

Implemented a fully automated karaoke server that allows users to queue songs via QR code, displays lyrics, and automatically classifies and renames new songs by language using local AI models, reducing manual file organization.

Рабочий процесс

• 1. Install PiKaraoke in a Linux container on a fast node. 2. Point PiKaraoke to the NAS library. 3. Configure PiKaraoke to display lyrics in the browser. 4. Generate a QR code on the TV for phone queuing. 5. Set up a nightly job that scans the top folder for new files. 6. Run two local language‑identification models on each new file. 7. If both models agree, rename the file to a standardized name and move it to the library. 8. If models disagree on Mandarin vs. Cantonese, run a third model (Jev) to decide. 9. If models cannot agree, leave the file pending for human review.

Доказательства

The plan had been obvious for a while. Demote the machine to what it’s good at, speakers and mics, and put the brains somewhere else.

Источник

The agent set up monitoring, which meant picking a search to run every hour as a health check, and it tried the searches this house would really make.

Источник

If they agree, it’s renamed and filed. If they both say Chinese but can’t agree which kind, a third model called Jev settles it.

Источник