Audience intelligence for catalogues, rosters and release decisions.
SoundIntel simulates how distinct audience segments respond to a track, from audio alone, before release. Built for the people who decide what gets signed, pushed and placed.
SoundIntel ranks unreleased tracks by predicted audience response, so your ears go where they matter.
Release prioritisation
Four finished tracks, one lead single.
Which one carries the campaign?
SoundIntel shows which track converts the broadest audience, and which one burns out fast, before the budget is committed.
Sync & placement fit
A brief asks for “warm, confident, youthful”.
Which tracks actually land that way?
SoundIntel maps which catalogue tracks actually land that way with the target demographic, instead of guessing from tags.
How pilots work
Result-only.
You send tracks. You receive reports. Nothing else moves.
No access to the engine, models or methodology
NDA-protected, unreleased and anonymised material accepted
Secure transfer, limited retention, deletion after pilot completion
Typical pilot: 50–200 tracks over 2–4 weeks
Proof
Blind. Timestamped. Public.
8 / 10
Qualifiers identified — Semi-final 1
7 / 10
Qualifiers identified — Semi-final 2
In a blind, timestamped test before the Eurovision 2026 semi-finals, SoundIntel identified 8 of 10 qualifiers in the first semi-final and 7 of 10 in the second, from audio alone, with no metadata, artist names or betting odds. It flagged Belgium, written off by the market at 13th of 15, as a sleeper. Belgium qualified. It read Bulgaria as a winner-compatible mainstream signal. Bulgaria won.