Fighting sound mismatches (2/2)
Since the May 2026 relaunch, some projects have sounded different for two reasons:
- Parts of the new audio engine behaved differently from the legacy engine.
- The conversion of legacy projects missed or changed some information.
Part 1 covered problem A, the audio engine.
In this update, we reconverted all eligible projects again using an improved converter, reconverted all existing presets, and added back the presets previously considered lost.
Here’s what changed, where to find your saved edits, and which sound differences may remain.
What happened to your projects
For projects created in the legacy Studio, reconversion depended on what you had done since the May 2026 relaunch:
| What you did since the relaunch | What happened |
|---|---|
| Edited and published the project | The project was left untouched. |
| Edited the project, but did not publish it after the relaunch | We saved a copy named [Before Fix] <project name> containing your post-relaunch edits. We then replaced the original project’s contents with the corrected conversion of its last saved legacy version. |
| Made no changes to the project | The project was updated with the corrected conversion of its last saved legacy version. |
The source for reconversion was your last saved version in the legacy Studio, before the relaunch. If you continued editing after publishing a track in the legacy Studio, those later saves are included. The reconverted project may therefore differ from the version you originally published.
Finding and choosing between versions
If we created a copy of your project:
- Open your projects view.
- Search for your project name.
You now have two versions:
- Your Project: your last saved legacy version, converted again with the fixes described below.
- [Before Fix] Your Project: your project as it was immediately before reconversion, including your saved edits since the relaunch.
Your post-relaunch edits are preserved in the copy. They have not been merged into the corrected original.
Compare both versions before deciding how to continue. You can keep working in the copy and publish it as a new track, or manually recreate or transfer the edits you want to keep into the corrected original. You can also keep both versions.
If you no longer need the copy, you can delete it after checking that it contains nothing you want to keep.
Legacy presets are back
Legacy presets that were missing have been added back to the library. Legacy presets already in the library have also been reconverted to bring their sound closer to the originals.
Presets created since the relaunch remain untouched.
What the updated converter fixes
The converter now preserves or translates project data that the initial conversion handled incorrectly:
- Orbit (formerly Graphical EQ): Corrected stereo separation automation. Frequency automation has also been rescaled so filter sweeps follow the legacy behavior more closely.
- Curve: Corrected the swapped automation mappings for the low-shelf and high-shelf filter controls.
- Matrix Arpeggiator: Restored missing note-input cables and pattern regions.
- Tonematrix: Restored missing groove assignments.
- Rasselbock: Corrected gate behavior at minimum and maximum widths. Gates at minimum width previously failed to open, silencing the row; gates at maximum width failed to close.
- Band Splitter, Beatbox 8, Crusher, Stompbox Compressor, and Stompbox Gate: Corrected parameter and automation mappings to follow legacy behavior more closely.
- Extreme groove settings: Groove settings that could not be mapped directly to the new engine have been converted into sample playback automation to preserve their original timing.
Legacy playback options
We’ve added back the legacy behavior of a few devices. You can find these settings in their context menus.
Pulverisateur
Right-click the device and choose Version → Flash to use the legacy sound model. Analogue is the default sound model.
Stompbox Reverb
Right-click the device and choose Version → Flash to use the original reverb algorithm. Freeverb is the default algorithm.
Groove on audio tracks
The legacy engine applied groove by changing audio playback speed in steps, once per sixteenth note. The new engine adjusts playback speed continuously. This can change how an audio region sounds when groove is applied.
To use the legacy behavior, right-click an audio region on a track with groove and choose Groove Mode → Quantized. Choose Continuous for the newer behavior.
Why some projects may still sound different
Our goal is to reproduce the legacy sound as closely as possible. Some differences remain because the new engine intentionally handles certain things differently:
- Region boundaries: The new engine enforces the end points of note and pattern regions. Projects that relied on playback continuing beyond those boundaries may sound different.
- Voice limits: Pulverisateur, Space, and Heisenberg can play more notes at once. In dense passages, notes that the old voice limits cut short may now continue sounding.
- Glide and slide: Pitch slides now behave consistently across devices. This can change passages that relied on a device’s previous slide behavior.
Complex effect chains or unusual routing may also still produce differences. If you hear something unexpected, please report it using the links below.
Compare the results yourself
We checked conversion accuracy by comparing audio rendered by the legacy and updated engines. You can explore results for more than 130 test tracks on our Conversion Mismatches test page.
The page lets you listen to the renders and compare three stacked spectrograms—images showing the sound’s frequencies over time:
- Top — Legacy: audio rendered by the legacy engine.
- Middle — New engine: audio rendered by the updated engine.
- Bottom — Difference: the differences between the two renders. Black areas show little or no difference at the displayed scale; colored areas highlight differences.
The visual comparison helps us locate differences; listening helps us judge their effect on the sound. The page contains a selection of test tracks, so your project may not be listed.
Still hearing a mismatch?
If your project sounds different and you’re unsure how to fix it, reach out.
Report it on Discord or through the bug report form. Please include:
- A link to the project, and whether you are checking the corrected original or a [Before Fix] copy.
- The timestamp or bar where the difference occurs.
- What you expected to hear and what you hear instead.
- An earlier audio render, if you have one, to help us compare.
Thank you for the reports, examples, and patience that helped us work through these fixes.