How it works
From a rough bounce to a ranked fix list in about a minute
Sonar measures your audio first and forms an opinion second. This page walks through every stage, what each score actually means, and the target each one is checked against.
Watch the 60-second tour
Prefer to read? Everything in the video is written out below.
The five steps
1. Upload a track
Drop an MP3, WAV, M4A, FLAC or OGG file, or a short snippet. A 60–90 second excerpt around the hook is usually enough for an accurate read. No account is needed for your first analysis.
2. The audio is measured
Before any AI opinion is formed, the file is measured: integrated loudness, true peak, dynamic range, spectral balance across bands, tempo, key and section boundaries. These numbers are facts, not guesses.
3. The song is scored
Those measurements plus the musical content are scored 0–100 across six dimensions. Each score comes with the reasoning behind it, so you can disagree with it for a stated reason.
4. You get a fix list
The report ends with ordered, specific actions — trim eight bars off the intro, pull 300 Hz on the vocal bus, lift the second chorus. Highest-impact fixes first.
5. Re-check the revision
Run the new bounce through and compare the two reports side by side. Your history shows whether the change actually moved the number you were chasing.
What each score measures
| Score | What it measures | Target we check against |
|---|---|---|
| Hook strength | How memorable and repeatable the main idea is, how early it arrives, and how often it returns. | Hook audible inside the first 20 seconds; returns at least three times. |
| Mix clarity | Separation between elements, low-mid build-up, harshness in the 2–5 kHz range, and headroom. | No single band more than ~6 dB above the genre reference curve. |
| Vocal balance | Vocal presence and intelligibility against the instrumental bed. | Lead vocal sitting roughly 3–6 dB above the busiest supporting element. |
| Arrangement | Structure, pacing, transitions and dynamic contrast across the full length. | A clear lift into the second chorus; no section repeating unchanged more than twice. |
| Streaming readiness | First-30-seconds retention risk and platform loudness normalization behaviour. | -14 LUFS integrated, true peak at or below -1 dBTP. |
| Commercial appeal | Sync potential, playlist fit and how current the production reads. | Reference-matched against recent releases in the same lane. |
Files, privacy and turnaround
- Accepted formats: MP3, WAV, M4A, FLAC, OGG — up to roughly 10 minutes.
- Turnaround: typically 30–60 seconds from upload to report.
- Storage: audio is analyzed and discarded, never kept or used to train anything.
- History: free reports save in your browser; members get cloud-synced history and revision comparisons.
Try the free tools while you are here
Common questions
- How long does an analysis take?
- Usually under a minute. Longer files take slightly longer because the measurement pass runs over the whole track.
- What file types can I upload?
- MP3, WAV, M4A, FLAC and OGG. A rough bounce is fine — you do not need a master to get a useful read.
- Do you keep my audio?
- No. The file is analyzed and discarded. Only the resulting report is saved, and on the free plan that report lives in your own browser.
- Is a snippet as good as the full song?
- For hook strength and mix clarity, yes. For arrangement and streaming readiness, upload the full track — those scores depend on how the song develops.
- Will two uploads of the same song give the same score?
- The measured values are deterministic. The scores move by a point or two at most between runs; a change larger than that reflects a real difference in the audio.
- Do I need an account?
- Not for your first analysis. An account saves your history to the cloud and lets you compare revisions over time.