FOSVED
← All projects

Alpha

HUK Radio

A free, non-commercial 24/7 international internet radio for music authors publish themselves

Repository on GitHub ↗

2 of 4 done · In progress: Interface and closed trial

Screenshots and mockups

  • Interface mockup (a design, not the finished app): radio, charts and track page on a phone
    Interface mockup (a design, not the finished app): radio, charts and track page on a phone
  • Interface mockup: playlists, an account with data export and deletion, the artist cabinet
    Interface mockup: playlists, an account with data export and deletion, the artist cabinet
  • Desktop interface mockup: the live radio with the player, the next track and the weekly chart
    Desktop interface mockup: the live radio with the player, the next track and the weekly chart
  • Desktop interface mockup: the weekly chart with language and style filters
    Desktop interface mockup: the weekly chart with language and style filters

HUK is a free, non-commercial internet radio that plays independent authors' music around the clock. All listeners hear the same moment of the broadcast. Authors keep their files: the platform stores only links, metadata and checksums, and holds audio just in a short-lived cache for what is currently on air. Charts (top-100 by language, style and direction) are computed by open, deterministic rules with no manual boosting. Moderation is narrow and aimed at clear violations of the law; AI triages the queue and humans decide hard cases.

What exists

  • A single shared broadcast for every listener, playback with a locked screen (PWA)
  • Sign-in by emailed link, author sign-up by invitation
  • Track submission by link, source verification and an automated moderation pipeline
  • Public likes and private dislikes, playlists, pre-moderated comments
  • Server-side verification of real listening and protection of the ranking from manipulation
  • Weekly category charts with archived snapshots
  • A short-lived audio cache for the current track and the next five
  • Export and deletion of your own data, an audit log, Russian and English interface

Highlights

  • Authors keep their files; the platform never stores third-party audio permanently, and CI verifies it
  • A transparent deterministic ranking and a set of security invariants that CI checks automatically
  • Built with AI agents — an agent executes tasks, a separate AI auditor reviews, a human decides

Not there yet

  • A public launch and a closed trial with invited people
  • Some interface screens from the approved design
  • Legal texts
  • A messenger bot for uploading tracks

Progress

  1. Foundation, security, moderation pipelineDone
  2. Listening, reactions, playlists, comments, charts, broadcast cacheDone
  3. Interface and closed trialIn progress
  4. Public launchPlanned

Next step: Deploy to a server and run a closed trial with a few invited people.

Stack

  • TypeScript
  • Python
  • SQL
  • Next.js 16
  • React 19
  • Tailwind CSS 4
  • next-intl
  • Auth.js
  • Prisma
  • zod
  • Vitest
  • fast-check
  • PostgreSQL 16
  • Docker Compose
  • Cloudflare
  • GitHub Actions
  • Claude-family models for moderation triage
  • audio fingerprinting services