Every feature is built around measured audio truth.

SongAnalyzerPro is not a prompt that guesses about songs. It is a ChatGPT app connected to a real MIR analyzer that reports what it can measure and labels uncertainty where it matters.

SongAnalyzerPro result interface showing analysis cards
01Tempo and beat gridMeasures global BPM, beat positions, tempo movement, and beat-grid structure for cueing, editing, reviewing, and arrangement work.
02Key detectionEstimates key with confidence and alternatives so ambiguous songs are handled as ambiguous rather than overclaimed.
03Energy flowMaps energy changes over time to expose intros, lifts, drops, breakdowns, and sections that deserve closer listening.
04Downbeat trackingUses dedicated downbeat analysis in strict mode. If downbeats cannot be measured, the app reports that instead of fabricating bar lines.
05Genre only when configuredGenre is classifier output, not a physical measurement. The app only reports genre when a validated classifier and taxonomy are configured.
06Provenance-first outputResults include method context so ChatGPT can explain what was measured, what was classified, and what should be treated cautiously.

The processing path.

Each step is designed to keep the user loop clear: choose a song, run analysis, inspect the measured result, and keep ownership and retention boundaries explicit.