On 2 July, Steinberg released SpectraLayers 13, the latest iteration of its spectral editor, once again available in Pro and Elements versions. Behind the in-house tagline “a new era of unmixing and spectral editing”, this version focuses on two specific areas: ambience isolation and voice processing. For those who do not use it every day, remember that SpectraLayers displays a picture of sound frequency by frequency, where you can paint, erase and move sound events like layers. The engine integrates directly into Nuendo, Cubase or any ARA host, and is just as useful for restoration as for sound design or remixing.
Unmixing finally tackles ambiences
The big idea in this v13 is called Unmix Sound Effects. The module separates short sounds – an impact, a footstep, a door, a hit – from the continuous background surrounding them: wind, crowd, traffic, natural beds. Each family goes to its own layer, ready to be replayed, attenuated or exported separately. Up to now, extracting a clean sound effect from a field recording required painstaking work with the spectral brush; turning it into a one-click process changes the game for sound editing.
The essential companion to this operation is called Ambience Heal. Any extraction leaves a hole in the spectrum, an area where the background has been torn away along with the event. Ambience Heal recomposes this missing material from the surrounding context, avoiding the infamous “holes in the air” and bubbling artefacts produced by overly aggressive denoisers. In my view, this is the most useful addition in this version.
Voice, the core of this version
Three new processes target speech, and they are the ones that will most interest dialogue, dubbing and podcast studios. Unmix Two Voices separates two speakers without having to record a profile for each one beforehand, whereas previous generations required training. Voice DeCrosstalk goes further: it removes an unwanted, non-targeted voice – typically spill from a nearby mic – even when it overlaps the main voice. Finally, Voice DeClick hunts down mouth noises, lip smacks and tongue clicks that contaminate close-miked takes.
Reconstruct: resynthesise rather than mask
The new Reconstruct module takes a different approach. Instead of copying or attenuating a damaged area, it resynthesises it from what surrounds it. The selection before and after the defect is used to rebuild sustained harmonics; the material above and below is used to rebuild transients and noise. Each of the three components can be adjusted independently, which allows you to make a “click” or a localised saturation strictly inaudible without carving a hole in the signal. For archive restoration, this is exactly the kind of tool that avoids sacrificing the life of the sound in the name of cleanliness.
Post-production: Atmos, Ambisonics and Pro Tools bridge
Steinberg has also strengthened the workflow side, the part that determines whether a tool will be adopted in a studio. The SpectraLayers Bridge for Pro Tools, via AudioSuite, now handles round-trip editing of multiple clips at once, with support for Dolby Atmos up to 9.1.6 and Ambisonics. Those used to immersive sound editing will appreciate no longer having to tinker clip by clip.
The remaining additions are all about productivity:
- batch processing applied to multiple open projects simultaneously;
- loudness measurements (True Peak, LU units) directly in the editor;
- Paste Into Selection and Paste Insert operations, time and frequency fades on selections;
- saving and recalling interface layouts, assignable keyboard shortcuts.
The new modules at a glance
| Module | Function | Use case |
|---|---|---|
| Unmix Sound Effects | Isolates a short sound from a continuous background | Sound editing, sound design |
| Ambience Heal | Rebuilds the spectrum after extraction | Restoration, cleaning |
| Unmix Two Voices | Separates two voices without prior profiling | Dubbing, interview |
| Voice DeCrosstalk | Removes an overlapping unwanted voice | On-set recording, podcast |
| Voice DeClick | Removes clicks and mouth noises | Voice, vocals |
| Reconstruct | Resynthesises the selected area | Detailed repair |
My take. I remember the days, working as a studio technical director, when recovering a voice buried in an already printed mix was pure fantasy: you either replayed it or gave up. Source separation has moved that limit into the realm of the possible, and SpectraLayers 13 confirms that working in the box is now fully viable even for these supposedly impossible jobs. I remain clear-headed though: aggressive resynthesis leaves micro-artefacts that the ear will always end up spotting, especially on sustained vocal lines. The real gain is not the party trick on an old master, it is the time saved in post-production and restoration. On that front, v13 hits the mark.
Pricing, upgrades and positioning
SpectraLayers Pro 13 is priced at €359 (£ pricing will vary), with a lighter Elements edition rounding out the range for occasional needs. Upgrade pricing is available for previous versions, and the update is free for those who activated v12 after 3 June 2026. On the system side, you will need macOS 12 or later and Windows 10 or later.
The competitive landscape has thickened in recent days. SpectraLayers moves forward against iZotope RX, which has just come under the Boris FX umbrella, while Waves’ real-time stem separation pushes the same technology towards mixing rather than repair. Steinberg, for its part, is aligning its range around a coherent immersive logic, in the wake of WaveLab 13 and its Dolby Atmos mastering. The full spec sheet and detailed system requirements can be found on the official Steinberg page.
