How to rebalance cleaned stems before the final mix

With AI artifact removal enabled, whole-mix cleanup runs before Enhanced restoration. The restored mix is then separated for targeted stem cleanup. Advanced mode lets you pause here to audition and adjust the stems before rebuilding the mix. Turn it on before upload; it starts off for each new submission.

This is mix control, not a miniature DAW. There are four relative-level faders, three vocal treatments, optional backing-vocal control, real-time reverb, a vocal harmonizer and three stem effects: a vocal de-esser, drum saturation and bass harmonics. You can correct a vocal that sits too far forward, lift drums that lost impact, tuck the other stem, or add a small amount of space around the vocal without opening a full production session.

Rebalance and Stem separation are different workflows

Rebalance lives inside an Enhanced job with AI artifact removal. Its working stems are temporary and are summed back into one finished mix. You do not download those stems.

If you need separate vocals, drums, bass, and other WAV files for a DAW, remix, karaoke version, or sampling session, use the dedicated Stems page. Our stem-separation guide explains the available engines and their limits.

1. Select AI artifact removal and Advanced mode

Start an Enhanced refinement, turn on AI artifact removal, then enable Advanced mode. The cleaner prepares the four working stems and pauses before the final mix. Rebalance is available on Pro, Studio and active welcome trials and adds 1 credit to Enhanced plus AI artifact removal: 4 credits total, or 5 with Finalize mastering.

Advanced mode starts off for every new submission and is not remembered as a preference. Turn it on deliberately for each batch you want to pause.

With Advanced mode off, processing skips the stem mixer. If Finalize is selected, you still choose Analogue Colour and reach final listening.

2. Preview the stems as one mix

Choose Preview mix and listen to vocals, drums, bass, and other together. The preview is an audition of the current controls, not a downloadable final render. Move one fader at a time and listen in context rather than judging an isolated stem.

The browser audition is approximate. The final render starts from the lossless stem checkpoints and uses server-side output protection, so its level and dynamics can differ slightly. Analogue Colour and mastering come later and are not included in this preview.

Each level is independent: raising drums does not intentionally turn every other stem down. The four stems still have to fit inside one output, however, so positive gains can trigger output protection to preserve headroom. Judge the relative balance, not just the loudness of one slider movement.

3. Choose the vocal treatment

• Original vocal stem is the default. Keep the stereo vocal and its existing ambience. This is useful when separation or reverb reduction changes a heavily processed voice too much.
• Process full vocal stem reduces existing reverb on the whole vocal before adding new reverb. Lead and backing stay together, so there is no separate backing-level control.
• Separate lead and backing uses a centered mono lead with reduced reverb. Adjust the backing-vocal level independently; new reverb goes to the lead, while backing vocals keep their sound.

The Vocals fader controls the whole vocal group. Compare all three treatments with your track: a heavily processed vocal may work better as the original stem than as a separated lead and backing.

Vocal noise cleanup is experimental and off by default. It applies to the selected vocal before new reverb. In Separate lead and backing mode, it affects only the lead. It can alter texture or remove parts of an unusual voice, so compare on and off.

Vocal reverb works with all three treatments and starts at zero. Choose a style from Bricasti M7 captures and adjust its amount, pre-delay, Low cut and High cut while listening. The browser previews these changes in real time; the final mix is rendered from the saved stems and your settings. Use Listen to reverb only to hear the wet return; at zero reverb, this is silent.

4. Shape the wet return with Low cut and High cut

Low cut removes low frequencies from the wet reverb return. Raise it when the tail adds rumble or low-mid buildup. Its minimum position is Off.

High cut softens high frequencies in the wet return. Lower it when the tail feels too bright, splashy, or separate from the track. Both filters affect the reverb only; the dry vocal stays unchanged.

5. Add the harmonizer and stem effects

Each effect has one amount control that starts at 0% (off). You hear changes live without playback restarting, and the saved mix uses the same processing.

• Vocal harmonizer sits beside the reverb. It uses the classic studio harmonizer technique, two slightly detuned copies panned hard left and right, to widen the vocal. The dry vocal is unchanged. A job prepared before the harmonizer was added may not offer it. The other three effects don't depend on when the job was prepared.
• De-esser softens harsh “s” sounds across the whole vocal, including its reverb and harmonizer. It finds each “s” and turns down only the part of the treble it occupies, so there is no voice setting. Mixes saved with the earlier de-esser keep their Low voice or High voice choice.
• Saturation on the drums adds Tape punch (odd harmonics) or Tube warmth (even harmonics). It works below 5 kHz, so cymbals stay clean.
• Harmonics on the bass adds upper harmonics so the bass is heard on phones and small speakers, where the lowest notes are often lost.

Reset returns every effect to off along with the faders and reverb.

6. Continue to Analogue Colour and final listening

Choose Continue to Analogue Colour to accept the stem and vocal settings. Colour is optional and processes this accepted mix. At final listening, choose your reference and mastering controls.

The initial listening preview uses no download slot. Choose Render & download to keep a version, then Confirm and close session when you have finished comparing. Confirmation keeps your main result and downloaded versions for 7 days and clears other previews and temporary working stems.

Closing the mixer leaves the job paused so you can reopen it from your dashboard. Closing final listening also leaves the session available: use Open final listening on the track to return. Generated previews are saved; unrendered control changes reset on close. Any render already started continues.

A restrained starting point

Begin with every stem at 0 dB and reverb and effects off. Fix the largest balance problem first, then make smaller moves. Add reverb and the harmonizer last, check the Low cut for buildup and the High cut for excessive brightness, use the de-esser if the added space makes “s” sounds harsh, and compare against Reset before rendering. Small changes are usually enough.

Rebalance is available on Pro, Studio and active welcome trials for 1 additional credit. The full workflow and current plan comparison are available on the help page and pricing page. Start free when you are ready to refine a track.