Smart Auto:
Audio Dominance
Detection.

01 — Abstract

What This Is

This document discloses a novel method implemented in IKANDY — a desktop music visualizer for Windows — for automatically detecting which audio-producing process on a host system is the dominant source, and silently switching the application's metadata retrieval backend to the most appropriate provider for that process. The method, referred to internally as Smart Auto, combines per-process audio session state inspection with backend arbitration, hysteresis dwell timing, candidate persistence across brief silent passages, and a denylist gate that prevents communication and broadcasting applications from incorrectly being promoted as the music source.

02 — Problem

The Listening Environment
Is Rarely Uniform

Music visualizer applications that display track metadata (title, artist, album art, synced lyrics) face a fundamental challenge: a user's listening environment is rarely uniform within a session. A user may listen via Spotify, then switch to VLC for a local file, then open a browser to stream a DJ set. Each source exposes metadata through a different mechanism with different limitations:

Prior art (Windows SMTC widgets, last.fm scrobblers, unified Now Playing daemons) reads from a single metadata source or polls all sources indiscriminately with no awareness of which process is currently producing audio. No existing system arbitrates metadata backends based on per-process audio production state.

03 — Method

The Novel Method

Smart Auto operates through six discrete steps, each adding a layer of robustness against the failure modes of naïve dominance detection.

// Simplified dominance evaluation
function getDominantProcess(sessions) {
  const active = sessions.filter(s => s.state === 'Active');
  if (active.length === 0) return null;
  const filtered = active.filter(s => !IGNORE_LIST.has(s.name));
  return filtered.sort(byAllowlistPriority)[0] ?? null;
}

// Hysteresis dwell guard
let candidate = null;
let candidateSince = 0;
const DWELL_MS = 1500;

function evaluateSwitch(dominant, now) {
  if (dominant?.name !== candidate?.name) {
    candidate = dominant;
    candidateSince = now;
    return;
  }
  if (candidate && (now - candidateSince) >= DWELL_MS) {
    commitBackendSwitch(candidate);
  }
}

Tier 2 SMTC Discovery

A parallel mechanism queries Windows' GlobalSystemMediaTransportControlsSessionManager via PowerShell to discover applications that register with SMTC but are not in the static allowlist (for example, Plex, Kodi, Apple Music, and various streaming clients). When such an application is detected as the active SMTC source AND a corresponding process is running on the host, it bypasses the allowlist gate and is eligible for dominance promotion. This provides extensibility without requiring an exhaustive allowlist of every media-aware application on Windows.

Backend Pause on Switch

When Smart Auto commits a backend switch, it issues a fire-and-forget pause command to the previous backend via that backend's native control API: Spotify's PUT /me/player/pause, VLC's ?command=pl_pause, or foobar2000's Beefweb POST /api/player/pause. This prevents the previous source from continuing to produce audio simultaneously with the newly-selected source. The previous-backend pause is fire-and-forget — its failure (e.g., Spotify Web API returning 404 when IKANDY is not registered as a Spotify Connect device) does not block the switch.

04 — Novelty

What Makes This Novel

The combination of these elements constitutes the claimed novel method:

05 — Implementation

Implementation Context

Smart Auto is implemented in IKANDY, an Electron-based music visualizer running on Windows. The audio session enumeration layer uses the electron-loopback-audio package, which provides a Node.js binding to the Windows Core Audio API. The metadata backend layer communicates via IPC between the Electron main process and the renderer. The visualizer itself renders via Butterchurn, a WebGL implementation of the MilkDrop preset format.

The Tier 2 SMTC discovery layer invokes PowerShell to query GlobalSystemMediaTransportControlsSessionManager, parses the resulting JSON, and cross-references the reported application user model IDs against the running process table.

Source code is maintained at github.com/IKANDYapp/IKANDY. Commit history provides timestamped evidence of invention predating this publication.

06 — Prior Art Search

Prior Art Search Summary

A search of npm, GitHub, and general web sources conducted May 2026 found no implementation combining the elements listed in §04 above. Existing packages such as electron-loopback-audio address system-level audio capture but do not perform process-aware metadata routing. Windows SMTC provides a unified metadata surface but does not select backends based on process identity, audio session state, or hysteresis-governed dominance.

Last.fm scrobblers and similar daemons consume metadata from a single source per session and require manual configuration to switch sources. They do not perform automatic arbitration based on audio production state.