Should Audio reviews include DSP?

[Mort]

If DSP is part of the product

Even that's complex with GLM. In the US, it's an add on, but in EU it's part of a monitor package.

 

[Arindal]

Audio products should be reviewed out of the box. Improvements to the product separately.

As mentioned, some loudspeakers come with built-in DSP which cannot be deactivated. Some (admittingly more affordable) speakers even come with continuous DSP-based room correction (Sonos comes to mind), or require room correction to function properly, or use DSP for continuous dynamic manipulation (such as SPL-dependent bass boost, limiter, loudness and alike).

The line between built-in DSP and separate DSP controllers or modules, is also pretty unsharp. Why review one speaker with DSP and the other one without, although a similar amount of processing power and capabilities are offered?

If DSP is part of the product, then absolutely it should be reviewed. How do we know how sophisticated it is otherwise?

I agree, but that puts a huge fraction of loudspeakers reviews under question, does it not? If a loudspeaker with built-in DSP would be granted the chance to employ room correction and evade all potential room-induced problem in the reviewer´s listening room, would that not be a great advantage over speakers which are not capable of applying such correction?

Or other way ´round: Wouldn't it be a good idea to review *any* speaker, including passive ones, with maximum of external DSP applied to potentially maxed out sound quality, to have a common ground for reviewing? I would consider that fair, and for the average reader it would mean a more realistic judgment about the maximum of sound quality achievable under given circumstances.

There is an implication to this except from being time-consuming and demanding extensive experience from the reviewer, though: anechoic frequency response measurements and subjective judgment of direct sound tonality as parts of the review would be rendered almost obsolete (if we exclude ridiculous flaws which cannot be properly equalized, such as resonance-induced issues or cancellation). Almost any curve can anyways achieved with the help of DSP, and availability of DSP is as cheap as it is easy nowadays.

 

[Sokel]

As mentioned, some loudspeakers come with built-in DSP which cannot be deactivated. Some (admittingly more affordable) speakers even come with continuous DSP-based room correction (Sonos comes to mind), or require room correction to function properly, or use DSP for continuous dynamic manipulation (such as SPL-dependent bass boost, limiter, loudness and alike).

The line between built-in DSP and separate DSP controllers or modules, is also pretty unsharp. Why review one speaker with DSP and the other one without, although a similar amount of processing power and capabilities are offered?



I agree, but that puts a huge fraction of loudspeakers reviews under question, does it not? If a loudspeaker with built-in DSP would be granted the chance to employ room correction and evade all potential room-induced problem in the reviewer´s listening room, would that not be a great advantage over speakers which are not capable of applying such correction?

Or other way ´round: Wouldn't it be a good idea to review *any* speaker, including passive ones, with maximum of external DSP applied to potentially maxed out sound quality, to have a common ground for reviewing? I would consider that fair, and for the average reader it would mean a more realistic judgment about the maximum of sound quality achievable under given circumstances.

There is an implication to this except from being time-consuming and demanding extensive experience from the reviewer, though: anechoic frequency response measurements and subjective judgment of direct sound tonality as parts of the review would be rendered almost obsolete (if we exclude ridiculous flaws which cannot be properly equalized, such as resonance-induced issues or cancellation). Almost any curve can anyways achieved with the help of DSP, and availability of DSP is as cheap as it is easy nowadays.

Applying the best case DSP to all speakers reviewed would make sense if they are about to be reviewed like Amir does with headphones.
I mean acoustically after the usual Klippel test.

That only happens to a few though, and on top of that the applied DSP should be tested prior to it, possibly comparing it with others so to know the best case one.

Seems enormous to me.

 

[90sNintendoKid]

I would say yes to include dsp as long as the reviewer states this and provides measurements pre and post EQ.

Otherwise in my opinion if a person is giving their subjective opinion on a speaker then it's how it sounds in their room.

For example, I have an untreated 3m by 2.1m room and my Kef Q350s have a peak around 50hz and to me they sound a bit boomy and muddy in the bass. I use EQ (Dirac ART and/or OCAs A1 Evo Acoustix) to help improve this.

I'd like to point out I EQ full range using Magic Beans True Target to calculate individual speaker target curves.

 

[voodooless]

While I do think that there is value in speaker reviews with these kinds of DSP options included, there are obviously some major caveats:

It's one thing to test simple setup changes, like low- and high-end offsets, or some simple EQ settings for certain physical setups. For these, it's obvious what they do, and easy to follow. And it doesn't make sense to review the speaker with these things set to default if that setup clearly does not match the physical setup of the DUT.

On the other hand, room correction is a tricky thing. It is working well, which is highly dependent on not just the technology, but also on the user, and most definitely on the room it is applied to. So whatever results you get, you won't necessarily get at home. But this is largely also true for a naked speaker, of course. It simply makes it really hard to test and compare. Still, the value of such built-in features does give it a major advantage vs a passive speaker, so it should be part of the whole review.

NO. Audio products should be reviewed out of the box. Improvements to the product separately. Cars shouldn't be tested with ECM hacks etc.

But there is no hack here... A car has loads of options that you can enable or disable. Do you think cars get reviewed without them? Should a 4x4 Jeep be tested in rough terrain without engaging the Four Wheel Drive? That doesn't make any sense...

How would one go about doing this kind of testing halfway objectively? I don't exactly know. There are so many variables involved that this is not at all trivial. And from that point of view, I understand why reviewers don't just do it. The practicality of it should be the primary discussion, I think.

 

[Arindal]

the applied DSP should be tested prior to it, possibly comparing it with others so to know the best case one.

Seems enormous to me.

Certainly true. It is not only time-consuming, but also both a reviewer and some sort of mastering engineer are mandatory, with lots of experience and a large library of different recordings for subjective tonal calibration. As existing recordings differ, to cover potential variety in tonality and a sufficient number of musical genres, I would put a guess of 500+ recordings which have to be reviewed for every single test, in order to have the correction curve really optimized for a broad range of applications.

the value of such built-in features does give it a major advantage vs a passive speaker,

And what about a passive speaker plus an amplifier or streamer having DSP capability? I mean, such are easily available nowadays, or even part of certain software like roon. If we count that in with active DSP speakers, maybe we should grant passive models the same opportunity.

How would one go about doing this kind of testing halfway objectively? I don't exactly know.

Maybe it would be a good idea to just admit that such level of objectivity in testing is simply not possible? In my understanding, that has an implication for any type of absolute verdict in a speaker review, particularly if referring to aspects which can easily be changed using DSP (such as the anechoic response).

 

[obsessive9119]

NO. Audio products should be reviewed out of the box. Improvements to the product separately. Cars shouldn't be tested with ECM hacks etc.

A better analogy is probably traction control, ie a function that tames the raw mechanical output in a way that makes it more pleasant. A car designed with TC in mind will be a very different experience when TC is disabled - and the same can be true of DSP.

 

[Sokel]

The hot potato here is not the speakers, is the DSP itself.
Market is flooded (and free stuff around also) and we pretty much take their word for it.

Dr Toole seems skeptical about the very ways they operate, I can quote posts like that.

I think it's time to find a way to take a good look under the hood, that's what meaningful at this time and age.
And I mean a complete break down, acoustically and electrically.

 

[KSTR]

DSP is a way too generic term and just means Digital Signal Processing.
It simply depends for what task any DSP functionality is used for, no matter if internal to the speaker or not, or from the manufacturer or not.

In this case, DSP is used for what is commonly labelled "room correction" which is always an optional task for speaker operation (I'm not aware of any speaker that cannot be operated without prior measurement to fit any processing to the true circumstances).

Therefore, the basic speaker test IMHO should be done without this feature but good practice would suggest that the feature is tested in a follow-up or two-part review, using the same setup at at the same time of course, for fair comparison of the feature on/off without any other variables changed.

 

[voodooless]

And what about a passive speaker plus an amplifier or streamer having DSP capability? I mean, such are easily available nowadays, or even part of certain software like roon. If we count that in with active DSP speakers, maybe we should grant passive models the same opportunity.

Maybe, and Amir often already does this by EQ'ing the speaker based on his measurements, and reports how it sounds with those corrections applied. But then the correction is not part of the product, and you'll have to buy it separately.

I think as a total value proposition, it makes sense: an active speaker with correction vs passive plus room correction amp is a very valid comparison that many consumers will make.

Maybe it would be a good idea to just admit that such level of objectivity in testing is simply not possible? In my understanding, that has an implication for any type of absolute verdict in a speaker review, particularly if referring to aspects which can easily be changed using DSP (such as the anechoic response).

I think a speaker should rarely have an absolute verdict. I see no point, given that you have so much data; condensing it to a single verdict is not very useful. Adding more data will not change this.

And if you only review it subjectively, it doesn't matter anyway, because it's subjective, anything goes.

 

[IsaacOscar]

Even for devices that are reasonable to use without EQ/DSP (e.g. a DAC) testing it would be helpful, for example here's a review that found some bugs in the Fiio Snowsky Melody PEQ, which would certainly annoy someone whose using the same EQ with another device/software, but would be hard for them to work out the problem. There were also similar bugs found for the TRN Black Pearl.

Just having a single EQ you test once whenever you review a product could reveal such issues.

 

[Drengur]

The obvious answer is "both" when possible. It also depends on what the reviewer is doing. Is it a product review? A technical review? It might be useful in many circumstances to know how a speaker behaves without optional components. What is the product description? Is it the manufacturers intention for people to always use the DSP? Can it be bypassed, upgraded, etc.?

I think the reviewer has to make a choice in each situation. Then it's up to the reviewer to be consistent and open about the choices.

More data = more better. I find it very helpful that Amir, for instance, tends to include optional EQ in his headphone reviews. It takes time and might not be the most common use-case, but it can say a lot about the product. I would never suggest he'd cut his reviews by x% in order to include measurements for every possible way of using a product.

Then you could of course ask; should reviewers measure modern subwoofers in such a way that they bypass the plate amplifier all together? It would be technically interesting, but not really relevant for most people.

I don't even know what I was trying to say.

 

[DoctorRobert]

I am highly interested in the performance of DSP products. I think that that is where the real improvements can be found these days.

However, for most I do not see how these can be objectively tested. A 'simple' speaker DSP, like the ones found on the back of active studio monitors to compensate for the location of the speaker, could still be relatively easily measured. But even in such a case, only the 'neutral' setting would have any meaning when comparing it to other speakers. If the for example the 'corner' setting is really the best choice for the corner in your room is a question that cannot be answered in a general test.

For products that include in-room measurements the situation is even more relative.

So for now, we know what we want from a stand-alone speaker, and it can be measured how well a speaker matches that demand. DSP products are best discussed in dedicated forum threads where experiences can be compared and knowledge about the application can be shared.

I still would like to see objectively comparisons of DSP products (but I would not know where to start).

 

[Purité Audio]

Even with traditional designs you can still use external EQ, reducing room gain ( boomy bass) will hugely improve the sound you hear.
Playback software, some dacs, streamers and even some contemporary loudspeakers now include EQ.

Keith

 

[Arindal]

(I'm not aware of any speaker that cannot be operated without prior measurement to fit any processing to the true circumstances)

Sonos Era series with its built-in Trueplay algorithm is an example.

the basic speaker test IMHO should be done without this feature but good practice would suggest that the feature is tested in a follow-up or two-part review, using the same setup at at the same time of course, for fair comparison of the feature on/off without any other variables changed.

Sounds like an option that is doing speakers both with or without DSP justice.

Regardless, I would expect a review with no EQ or room correction activated, to not include a verdict on aspects that depends vastly on individual EQ, such as tonal balance, and at least a minimum of investigation done by the reviewer, if any incompatibilities with the room or setup are in play that would strongly suggest employing EQ under field circumstances. Having particularly room modes and resulting boomy bass in mind, but also imaging and reflection issues.

A loudspeaker review to a certain degree is a subjective and individual thing, with the reviewer trying to meet the necessary conditions for the speaker in question. I mean, you would not review a sound reinforcement system under nearfield conditions or a nearfield monitor in free field conditions, or a bass-heavy system placed against the wall. Applying EQ or basic room correction, to me fall into the same category of trying to adopt the speaker to the given conditions before beginning the actual subjective evaluation.

I think a speaker should rarely have an absolute verdict. I see no point, given that you have so much data; condensing it to a single verdict is not very useful.

I fully agree, but I oftentimes read about speakers being reported to have certain tonal flaws as seen in their anechoic response, or alike. What is the point of criticizing such aspects if you can easily and quickly correct them? I would be more interested if they are actually correctable, or what is the limits/downsides of applying EQ to the speaker in question.

And if you only review it subjectively, it doesn't matter anyway, because it's subjective, anything goes.

I do think that subjective reviews are at least partly useful, if they contain A/B testing, predefined parameters how to evaluate sound quality, and if the findings can be applied to other cases as well, so a precise language is key here. If everything would be purely subjective and anything goes, well, I do not see a point in publishing anything at all.

Even with traditional designs you can still use external EQ, reducing room gain ( boomy bass) will hugely improve the sound you hear.

Interesting point. While you are right that external EQ can be applied on almost any speaker design, an important part of a speaker review in my understanding would be to explain the limits of potential EQing. In some cases, boomy bass can be corrected by DSP, in others they cannot to full satisfaction.

 

[KSTR]

Sonos Era series with its built-in Trueplay algorithm is an example.

From what I can find on the web nothing stops you to install and use these Sonos speakers with Trueplay turned off (there is switch in the mobile app).

 

[markstein]

Erin (Audio Corner) got a lot of flak when he reviewed the Genelec 8361a without GLM, the Genelec bass management, active crossover, and DSP system.

The Genelec army pointed out that GLM is made by the manufacturer and designed only to be used by Genelec SAM products. It should be tested as actually used.

Erin understood those points and responded with good points like doesn't that put passives at a disadvantage? And it takes up a lot more time to test DSP.

There's a lot more to say about it, obviously. I think Erin promised to include GLM if they send him an 8380 to review. I hope they do.

tl:dr When should DSP be a part of an audio measurement review?

As DSP can be done in a good and a bad way - result not related to the DUT - it should be done only as part of the subjective listening test.

Sound & Recording/ Anselm Goertz e.g. always applies a Room EQ when doing listening tests and shows the original mean frequency response (mean value of 30 measuring positions around listening location), the corrective filter and the resulting mean frequency response in his reviews.

AFAIK he does not use proprietary systems like MA1 or GLM but does all DSP corrections systematically comparable, using his own DSP hardware, leading to a somehow more "objective" subjective comparison.

 

[gwing]

Let me suggest an analogy:

Say you review cars and one of the review criteria you usually apply is the quality of it's steering response. You have two cars to compare , one has power assisted steering and the other does not - what do you do?

1. Get out your screwdriver and remove all power assistance, if any, from both cars before testing?
2. Design and fit power assiatance to the other car's steering?
3. Alter both cars to use some arbitrary but identical third party power assistance to both cars?
4. Test the cars as they are built and supplied as standard by the manufacturer?

 

[popej]

There is a better analogy: should an AVR be tested with DSP? Still the answer is not obvious. I would like to see tests with DSP.

 

[Sokel]

There is a better analogy: should an AVR be tested with DSP? Still the answer is not obvious. I would like to see tests with DSP.

Who on earth would use an AVR without DSP on?
It's like like getting a miniDSP just to use it as a DAC.

OBVIOUSLY they have to be tested as in real use, that's their purpose and that's why they have a category of their own.