This is the workflow TinyBooth is built around: Suno bundle → cleanup → mix → release. Each step uses the v0.4.0 Suno-aware features (per-role correction presets, import-time coherence verification, polarity flip, DC trim, Nyquist cleanup, BS.1770 LUFS metering, project trim) without asking you to know what BS.1770 means or which knob does what.
If you're new to TinyBooth, start with Importing Suno stems for the basics, then come back here for the full mixing flow. The Mix tab manual covers the moment-to-moment console layout in more detail.
Suno gives you two delivery shapes — a zipped "Download All" archive or a folder of individual stem WAVs. TinyBooth handles both via the File menu.
Pick the matching entry. The dialog asks you where to extract the project; the default is a sibling folder next to the source.
When the import finishes, a modal pops up with everything TinyBooth learned about the bundle:
-
Stem count and where each one landed.
-
Mixdown LUFS — the bundled mixdown's BS.1770-4 integrated loudness, computed once at import. This is your target loudness for the rest of the workflow.
-
Coherence verdict — sums every stem at unity, subtracts the mixdown, reports the residual relative to the mixdown's own RMS:
Residual relative to mixdown Verdict ≤ −30 dB Stems compose cleanly into the mixdown ≤ −20 dB Stems mostly compose — minor residual ≤ −10 dB Noticeable residual — a stem may be missing or anti-phase > −10 dB Stems do NOT compose into the mixdown -
Per-stem polarity check — Pearson correlation between each stem and the mixdown. Stems with
r < −0.3get an⚠ ANTI-PHASEbadge. Don't panic — these are easy to fix in step 4.
If anything in the report looks wrong (a stem flagged anti-phase, a weak coherence verdict), make a mental note. Then close the modal.
Switch to the Mix tab. The bundle is already loaded as a multitrack project with each stem on its own lane.
Each stem already has a correction chain attached — the appropriate Suno-X preset for its role:
| Role | Auto-seeded chain |
|---|---|
| Vocals | Suno-Vocal — HPF 90 Hz, mud cut, presence + air, de-esser, Nyquist clean |
| Backing Vocals | Suno-BackingVocal — tighter compression, lighter de-essing |
| Drums | Suno-Drums — no HPF (kick needs sub), box cut, stick attack, DC remove |
| Bass | Suno-Bass — HPF 30 Hz, mud scoop, slow-attack comp, DC remove |
| Electric Guitar | Suno-ElectricGuitar — low-mid cut, presence lift |
| Acoustic Guitar | Suno-AcousticGuitar — body + air, light compression |
| Keys | Suno-Keys — mud cut, presence lift |
| Synth / Lead | Suno-Synth — AI-shimmer notch at 14 kHz, Nyquist clean at 17 kHz |
| Pads | Suno-Pads — cloud cut |
| Percussion | Suno-Percussion — snap lift, air shelf |
| FX / Other | Suno-FxOther — light glue, conservative |
The defaults are tuned for typical Suno output. You'll tune them per-track in step 6.
If the import report flagged any stem as ⚠ ANTI-PHASE, click the
Ø button on that stem's channel strip. The button highlights
when the polarity flip is on; the audio inverts immediately on the
next playback cycle.
A polarity-flipped stem on its own sounds identical to its non-flipped counterpart — your ear can't hear absolute polarity. But when summed with the other stems, an inverted stem cancels rather than adds. Flipping it brings the cancellation back into alignment.
To verify: the project's tracks/ folder still contains the
original Suno mixdown (TinyBooth keeps it as a reference rather than
summing it as a track). You can solo your mix and the mixdown
side-by-side in any audio tool to A/B.
Suno songs sometimes have a few seconds of dead air at the start, a fade-out tail at the end, or a count-in you don't want. Open Project tab → ✂ Trim project….
The trim panel shows a waveform thumbnail of the bundled mixdown (or
the first track if no mixdown). Enter mm:ss.mmm start and end
times — the markers update live as you type, and the panel tells
you exactly how much it'll cut from each end.
Hit ✂ Apply trim. Every WAV in the project (every stem and the bundled mixdown) gets cropped to the same range, atomically. The trim is destructive — re-import the bundle if you need to recover the originals.
Coherence stays valid post-trim because every file shares the same new frame-0; the bundled mixdown gets the same crop as the stems.
The auto-seeded Suno-X presets are starting points, not finishing points. For each stem worth tuning:
- Solo the stem (the
Sbutton on its strip). - Click + Correction on the strip to open its chain editor.
- Adjust EQ bands, compressor, de-esser, gate, DC trim, Nyquist clean. Changes apply on the next playback cycle.
- Un-solo and listen in context.
A few rules-of-thumb worth keeping:
- Nyquist clean is on by default for every Suno-X preset. Turn it off only if you specifically want the top-octave shimmer.
- DC remove is on for drums / bass / percussion. It's cheap and reclaims a few millivolts of headroom; rarely worth turning off.
- The de-esser is on for vocals only by default. If a synth lead has harsh sibilance-like artefacts, enabling it briefly can help.
- The compressor makeup field is not an output trim; it's there to put back what the compressor takes off. If your mix is too quiet after compression, raise makeup, not the master fader.
For the per-tab UI details, see the Mix tab manual and the Admin / profile-editor manual.
The transport bar shows the master bus's loudness in real time:
- M — momentary (mean over the most recent 400 ms).
- I — gated integrated (whole-programme mean per BS.1770-4 §5.1 with the −70 LUFS absolute and −10 LU relative gates applied).
Reads — until 400 ms of audio have played. Resets on Stop.
Streaming-service targets:
| Target | LUFS |
|---|---|
| Spotify | −14 |
| Apple Music | −16 |
| Tidal | −14 |
| YouTube Music | −14 |
| Broadcast (EBU R128) | −23 |
Adjust the master fader to bring I toward your target. Reference
point: the bundled Suno mixdown's own LUFS was logged at import
time (open the import log file under
%APPDATA%\TinyBooth Sound Studio\logs\). Matching that loudness
makes your remix sit at the same level as the original Suno output.
Switch to the Export tab. Pick a format and a target file.
Native WAV export is built in. The lossy / FLAC formats route through
ffmpeg, which TinyBooth looks for next to the executable or on
your PATH (see the
Export manual if it isn't found).
The export sums every track through its correction chain at the master fader's static gain (automation lanes are honoured). No limiter is applied; if your mix peaks above 0 dBFS the export will clip — keep an eye on the master peak meter while you set the fader.
The v0.4.0 Suno-cleaning surface is intentionally focused. A few adjacent features are on the roadmap but not yet shipped:
- Reference playback A/B — toggle between your mix and the
bundled mixdown at matched loudness, single-button. The LUFS meter
- the import-time mixdown-LUFS reading are the foundation; the playback-source swap arrives in v0.5.0.
- Multi-take browser — load multiple Suno generations of the same song side-by-side. Borrow stems across takes. Multi-take A/B.
- Per-track non-destructive trims with drag handles on the Mix tab waveform lanes. The current trim panel is project-wide and destructive (rewrites WAVs). Per-track non-destructive trims are a v0.5.0 sequel.
The modules involved in this workflow:
| Module | Role |
|---|---|
src/suno_import.rs |
Bundle ingestion (folder + zip), role classification, mixdown detection |
src/coherence.rs |
Sum-vs-mixdown residual + Pearson correlation for the polarity check |
src/lufs.rs |
BS.1770-4 K-weighting + integrated loudness with gating |
src/dsp.rs |
The 11 per-role Suno-X presets + the chain that applies them |
src/trim.rs |
Project-level batch trim (.tmp + rename atomic writes) |
src/ui/mix.rs |
Mix tab — channel strips, waveform lanes, transport bar |
src/ui/correction.rs |
Per-track Correction window |
src/ui/trim.rs |
Trim panel |
For the cross-cutting view, see
docs/architecture.md.