Are DSP presets an inefficient way to handle small tonal preference changes?

[klettermann]

In DSP-based systems, presets are commonly suggested as the correct way to accommodate tonal variation. I’m questioning whether that’s actually good system design for small, reversible preference adjustments. DSP is exceptionally well suited to structural calibration — room correction, timing, phase alignment, and setting a reference target response. Once validated, that configuration ideally remains stable. DSP presets clearly have an important role. They are extremely useful for handling genuinely different system states, for example:
>optimizing for a single listening position vs a wider seating area
>nearfield vs farfield listening
>materially different spatial targets
Those are discrete use cases that require and justify separate measurement, validation, and recall.

However, presets are also discrete and relatively complicated to create and test (as many ASR threads attest). More importantly, a limited number of presets seem best treated as a small selection of system configurations rather than as variable controls. Using presets to approximate small, continuous tonal preferences (e.g. ±1–4 dB overall tilt or modest low-frequency lift) feels like a poor fit for the tool. It entangles calibration work with moment-to-moment listening preference and quickly prods one toward a preset-proliferation rabbit hole. In practice, the limit is simply the number of presets available.

From a system-architecture standpoint, this mixes two different functions:
>FR calibration and spatial optimization (structural, measurable, validated, stable)
> Preference adjustment / tone control (continuous, exploratory, reversible

A dedicated post-DSP trim stage (digital or analog) keeps these roles clearly separated:
> calibration and spatial optimization remain fixed
>preference adjustments are continuous and immediate
> changes are reversible and non-destructive
> no additional presets are consumed

The question I want to debate is not whether DSP can do this — it obviously can — but whether it should. Why is using finite, labor-intensive DSP presets for small tonal trim treated as best practice, rather than as a workaround for the absence of a proper trim stage? If the objection is determinism, recallability, or measurement purity, that’s a valid engineering preference. But it’s not obvious to me that those benefits outweigh the architectural and workflow costs of using presets for a role they’re not well matched to. I’m less interested in “either way is fine” answers than in arguments grounded in system design tradeoffs rather than personal preference. Thanks, Happy New Year and cheers,

 

[Oddball]

Dedicated room correction system will calculate its filters based on the calibration case. So it will provide best it can, including any manual adjustments to curves etc.

Altering that after the fact might have impact on the balance of the system as intended by the specific room correction system and the curve. That will be different based on what room correction system is in use and what is one correcting after the fact.

Lots of people seem to like having tone controls at the end of the processing chain. Good or bad to have/engage them? Probably good as usual number of presets on most gear might not be sufficient to adjust for preferences for variety of material people are listening to. But as noted, might throw of the base calibration as it was generally intended to be used without adjustments.

 

[Keith_W]

The question I want to debate is not whether DSP can do this — it obviously can — but whether it should. Why is using finite, labor-intensive DSP presets for small tonal trim treated as best practice, rather than as a workaround for the absence of a proper trim stage? If the objection is determinism, recallability, or measurement purity, that’s a valid engineering preference. But it’s not obvious to me that those benefits outweigh the architectural and workflow costs of using presets for a role they’re not well matched to. I’m less interested in “either way is fine” answers than in arguments grounded in system design tradeoffs rather than personal preference. Thanks, Happy New Year and cheers,

It is possible to obtain a DSP tone control by:

1. Baking it into your target curve, and generating multiple filters based on different target curves as you describe. Depending on your filter generation software, this may be onerous. I use Acourate - once the measurement is taken, it takes me 5 minutes to generate and load filters with a different target curve. If I were to do it with REW, it would take much longer since that part of the process is not automated.

2. Use a VST tone control - here are some options. Or use a convolver with built-in PEQ, like Acourate Convolver. JRiver has a built-in parametric equalizer, but it's not particularly easy to use.

 

[klettermann]

Thanks — this is a helpful framing, and I mostly agree with it.

I think the crux is what we mean by “throwing off” the calibration. If a post-DSP trim stage were altering time alignment, phase relationships, or spatial correction, I’d agree that it undermines the intent of the room correction system. But for small, broadband tonal adjustments (say ±1–2 dB tilt or a gentle LF shelf), the underlying structural correction remains intact. That’s really the distinction I’m trying to get at: separating structural validity from tonal balance. Room correction establishes the former; tone trim adjusts the latter. Those two layers don’t seem inherently incompatible unless we assume the calibrated target curve is itself sacrosanct rather than a reference. I’d also argue that this is precisely why presets work so well for different listening positions or spatial targets, but less well for small, continuous tonal preference changes. In the latter case, using presets feels more like compensating for the lack of a trim layer than an optimal use of the tool.

So I agree there’s a risk of undermining calibration — but only if calibration is defined as “the exact target curve must remain untouched,” rather than “the system’s structural corrections remain valid.” Curious how others draw that line.

 

[amirm]

That’s really the distinction I’m trying to get at: separating structural validity from tonal balance.

That is the process but there is no harm to then merge the two as suggested. When I correct for speaker and headphone response errors, I first apply corrections based on measurements and then adjust them for taste across a good set of music clips.

At the end, it is all a compromise since there are no standards in music so you can still get content that has too much bass, too little bass, too much highs, too little highs, etc. Fortunately your ears will adapt.

 

[middlemarch]

Problem is how to merge the two. Currently with, for example, Audyssey, there are tone controls as long as you don't use their Dynamic EQ. Which sort of makes sense, and DEQ is somewhat configurable to taste within limits. But with DIRAC, for example, no tone controls at all. You're limited to basically three target curves to switch between.

I think that's still the big unsolved problem, how to get the functionality of the old Bass and Treble knobs incorporated into the modern room correction systems. I've heard maybe Storm and Trinnov have some functionality in this regard? But don't know for sure, and too pricey for me.

 

[Keith_W]

I think that's still the big unsolved problem, how to get the functionality of the old Bass and Treble knobs incorporated into the modern room correction systems. I've heard maybe Storm and Trinnov have some functionality in this regard? But don't know for sure, and too pricey for me.

If you are using a PC as your source, you just stick something like this in your signal chain:

 

[respice finem]

I like the simple, yet fast and effective RME approach. I can choose from many EQ presets, but if I'm lazy (and I am, most of the time), I can adjust via old-fashioned B/T knobs (still in the digital domain).

 

[Oddball]

Problem is how to merge the two. Currently with, for example, Audyssey, there are tone controls as long as you don't use their Dynamic EQ. Which sort of makes sense, and DEQ is somewhat configurable to taste within limits. But with DIRAC, for example, no tone controls at all. You're limited to basically three target curves to switch between.

I think that's still the big unsolved problem, how to get the functionality of the old Bass and Treble knobs incorporated into the modern room correction systems. I've heard maybe Storm and Trinnov have some functionality in this regard? But don't know for sure, and too pricey for me.

D&M has tone controls for Audy as noted but not for Dirac. Hopefully that will come with some future firmware update. Storm has PEQ filters before Dirac and then tone controls after Dirac, so more flexible (and expensive).

Not sure what Trinnov has, but it is billed as most flexible of them all and would be surprised if it did not have tone controls. But also has mega number of presets so arguably one could use all these instead of tone controls.

An example of when you throw off the calibration with after-the-fact adjustments would be Dirac ART. Max bass boost for curves/shelves is +12dB, and for some people that is apparently not enough. One can increase gain on the subs (which would to some extent be similar to tone control) to gain some further bass impact, but at that point the effectiveness of ART, especially in decay control, starts to be degraded. That does not automatically mean that it is a bad thing. Decay with ART can be pretty low, and some associate that with "dry" sound. Increasing delay beyond ARTs calculations could be a matter of preference. However, at one point ART will stop being ART.

+1 or +2 dB would certainly keep ART effective, albeit a bit less, while going above +6dB increases delay significantly - at least in my system. This is also room/system specific as some get lower and some get higher decay with ART. Unfortunately cleaned up my ART folder as it was getting too big so no pretty graphs to show - but for whatever it is worth, been there measured that.

 

[sigbergaudio]

Problem is how to merge the two. Currently with, for example, Audyssey, there are tone controls as long as you don't use their Dynamic EQ. Which sort of makes sense, and DEQ is somewhat configurable to taste within limits. But with DIRAC, for example, no tone controls at all. You're limited to basically three target curves to switch between.

I think that's still the big unsolved problem, how to get the functionality of the old Bass and Treble knobs incorporated into the modern room correction systems. I've heard maybe Storm and Trinnov have some functionality in this regard? But don't know for sure, and too pricey for me.

The old controls can be emulated with regular PEQ. So I wouldn't say it's super-unsolved. But it was of course more practical to have an actual physical knob to adjust with.

 

[Eric Natural]

At the end, it is all a compromise since there are no standards in music so you can still get content that has too much bass, too little bass, too much highs, too little highs, etc. Fortunately your ears will adapt.

My definition of "no standards" is inside recordings. Almost no two sound alike - I feel like I'm hearing each studio much more than the gear . . .

 

[klettermann]

Thanks to all — this has been very illuminating. It’s probably time for me to take a step back and reflect on what led me to this topic in the first place.

I more or less stumbled into a solution that works well for me, though I was originally solving a completely different problem. Before going all-digital, I bought a Schiit Loki Max for the simple reason that I wanted tone controls. At the time, my AV and 2-channel systems were back-to-back in the same long room and shared nothing. A couple of years ago I moved and ended up with a much smaller room. The Loki turned out to be a very useful bridge: switchable RCAs for AV and XLRs for 2-channel audio meant I could share the same main speakers and amplifier, with simple tonal adjustment when needed. Not a perfect solution, perhaps, but a very practical one.

Then, about a year ago I went down the DSP rabbit hole — no more preamp, no more LPs, just a MiniDSP SHD Studio. After a lot of learning, fiddling, relearning, and more fiddling (thanks Keith!), I got it set up as shown in the diagram. The result sounds glorious — far greater than the sum of its parts. During all of that, the Loki mostly sat there as an expensive source switch. But after months of listening, I gradually started using it again for small tonal tweaks. Unsurprisingly, it’s vastly easier than creating and managing multiple fixed DSP presets.

That led to the obvious question: why can’t I do this just as cleanly in the digital domain? Of course you can, but in practice it feels awkward and over-engineered for such a simple task. Now this leads me to a broader conclusion: DSP and digital audio are still, in some important ways, in their infancy. The underlying mechanics are extremely powerful and work very well. The UI and UX however, especially for separating calibration from casual preference adjustments, lag far behind. It will be fascinating to see how the industry closes that gap over the next 5–10 years.

Happy New Year to all, and thanks again.

 

[middlemarch]

I guess my comments were kind of limited to the AVR world, which is what I use, so a bit parochial. Nice to know that in the two channel world there are better solutions. Still would love to see that functionality become general. I think you've summed it up nicely, it's easy to lose sight of how far we've come in many ways over the last decade or two. Still early days, so I'm sure the convenience factor will be taken care of once the really hard problems have been solved.