Albumtide mastering, as an API.
Send a mix, get back a master ready for every platform, with a report that explains each decision. The same engine as in Albumtide: signal processing, not a black box.
Processing in Switzerland · pay per use · first sign-ups get in first# one mix, one style, two loudness targets curl https://api.albumtide.com/v1/masters \ -H "Authorization: Bearer $ALBUMTIDE_KEY" \ -F file=@mix.wav -F style=deep \ -F targets=-14,-9
{ "id": "mst_8f2c", "status": "done",
"masters": [
{ "target": -14, "lufs": -14.0, "true_peak": -1.6 },
{ "target": -9, "lufs": -9.1, "true_peak": -1.1 } ],
"report": { "lowmid_db": -1.8, "deess_db": -0.7,
"limited_db": 2.4, "warnings": [] } }Eleven steps, in this order.
What the engine does to every mix. Defaults come from the chosen style; each optional module can be dosed or turned off.
- Analysisalways
1/12-octave spectrum measured on the loud passages, integrated loudness, true peak, stereo width per band.
- High-passalways
24 dB per octave, at 20-30 Hz depending on the style: removes inaudible sub-bass that eats the limiter's headroom.
- 5-filter EQadjustable
Only corrects the gap between the mix and the style's target (or the reference), both ways: ±3 dB at most. A mix already on target is left untouched.
- Dynamic low midsadjustable
Lowers 180-500 Hz only when that range builds up in the mix. The rest of the time, nothing changes.
- High-end softeneradjustable
Tames 6-12 kHz when the highs bite: steady harshness with a cut placed on the excess (6 dB at most), bursts as they come (common with generated sources); the air above stays intact. On by default only if the analysis finds harsh highs.
- Mono bassalways
Below 85 to 135 Hz depending on the style, the low end is centred: solid on a phone and in a club.
- Stereo widthadjustable
Widens or narrows above a chosen frequency; below it, nothing changes. A guard prevents phase cancellation.
- Saturationoptional
Soft harmonics, 4× oversampled to avoid aliasing.
- Multiband compressoroptional
Four bands, each with its own ratio, threshold and timing, blended with the original sound.
- Glueadjustable
Gentle compression, the style's own, set from the track's level, the same on left and right. None for Jazz / Classical and Ambient.
- True-peak limiteralways
Hits the requested loudness while measuring peaks between samples; release adapted to the track so the low end doesn't pump. 24-bit WAV out.
No saturation or multiband by default: those are the two modules that make a mix pump when they're overdone.
A master you can explain.
No trained model that changes without warning: signal-processing modules, dosed for each track and style.
The same file with the same settings gives the same master, today and a year from now. You can test it, reproduce it, document it.
Every master comes with its report: loudness, true peak, and what each module did, in dB. Your users know what changed.
A 4× oversampled limiter and headroom tuned for the AAC and Opus encoding platforms apply: no clipping after upload.
A target curve per style, or matching a reference track that is analysed then deleted: only its fingerprint is kept.
-14 LUFS for streaming, -11 for social, -9 for club: the chain is computed once, only the limiter changes.
Processing on our servers in Switzerland. Files never train anything and are only kept for as long as delivery takes.
For platforms that deliver music.
Pricing is announced at launch. People on the waitlist hear first and can test before everyone else.
Early access.
Tell us what you want to master and how much: we open the API first to the use cases that fit best.