Kallitone says blind listener tests ranked AI tracks below average
Kallitone says 25,000 blind music comparisons from a year-long beta showed all 40 AI-generated tracks in the test set scoring below the catalogue average. The results point to a discovery model that uses direct listener choices, not artist reputation, to surface unfamiliar music.
Why it matters: - Kallitone is testing whether large-scale blind listening can help unfamiliar music find an audience without artist names or existing reputation shaping the outcome. - The beta results suggest listener preference data can be turned into rankings for discovery, including a separate signal on AI-generated tracks in the sample.
What happened: - Kallitone published results from 25,000 blind music comparisons after a year of beta testing. - The beta covered 500 tracks: 460 from independent musicians and 40 identified as AI-generated. - Each track appeared in 100 completed comparisons, creating 50,000 track appearances across 25,000 paired voting decisions. - The public test used anonymous comparisons so listeners chose between recordings without seeing artist names.
The details: - Listeners heard randomly selected excerpts from within two recordings. - Each track had a minimum listening requirement of 15 seconds. - Excerpts were not automatically taken from the beginning of a song. - Artist identities were withheld during voting. - Completed votes fed into Elo-style ratings. - The system uses both comparison outcomes and the ratings of competing tracks, rather than simply sorting by total favorable votes. - The resulting Top 20 included tracks from 10 artist projects. - The Top 20 spanned electronic music, indie pop, piano ballads, rock, blues, grunge, acoustic music and reggae. - All 40 AI-generated tracks scored below the catalogue's reported average Elo rating of 1250. - The highest-rated AI-generated track scored 1244. - Kallitone said these results describe the beta repertoire and participants, not a general verdict on AI-generated music. - Kallitone's public listening page shows the current beta rankings.
Between the lines: - The beta is as much a test of discovery mechanics as a music ranking exercise. - The results point to a system where many small listener judgments can shape what rises, which may help surface music that would otherwise stay buried. - The AI-track results are limited to this sample and should not be treated as a broad market conclusion.
What's next: - Kallitone said further testing is needed to see how the approach performs with larger catalogues and audiences. - Rankings may change as more votes are added. - Kallitone plans to keep the beta open for visitors through its public listening page. - Kallitone is a music discovery platform based in Bucharest, Romania, operated by Orkaan Studio S.R.L.
Disclaimer: This article was produced by AGP Wire with the assistance of artificial intelligence based on original source content and has been refined to improve clarity, structure, and readability. This content is provided on an “as is” basis. While care has been taken in its preparation, it may contain inaccuracies or omissions, and readers should consult the original source and independently verify key information where appropriate. This content is for informational purposes only and does not constitute legal, financial, investment, or other professional advice.
Sign up for:
Music Industry Watch
The daily local news briefing you can trust. Every day. Subscribe now.
Check Your Email!
We sent a one-time activation link to: .
Confirm it's you by clicking the email link.
If the email is not in your inbox, check spam or try again.
Welcome back!
is already signed up. Check your inbox for updates.