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.

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.
Upload selection
The user selects an audio file from ChatGPT's file flow. The app receives a short-lived URL, not an instruction to keep a music library.
Request orchestration
The app validates requests, serves the ChatGPT widget, and forwards short-lived analysis work without treating submitted songs as stored library content.
MIR processing
The analyzer uses Python audio tooling such as ffmpeg, librosa, NumPy/SciPy, and downbeat analysis models where configured.
Structured result
The app returns tempo, key, energy, downbeats, genre status, confidence, and provenance as structured data for the UI and ChatGPT.
Deletion window
Songs are used only for analysis, deleted from servers within 24 hours, and never repurposed for training or unrelated use.