Axiom Home Page Forums General Discussion What's New At Axiom Broadband DSP Correction

The introduction of the LFR1100 brought about a lot of questions about using a Broadband Digital Signal Processor (DSP) between your pre-amp and your main amp in other speaker models for the purpose of smoothing out small anomalies in the amplitude response. These are referred to as ‘High Q’ corrections to the response curves.

In many ways a speaker’s amplitude response curve could be divided into two Q groups:
• The Low Q, which deals with amplitude differences over a wide frequency band
• And the High Q, which deals with amplitude differences within a narrow frequency band.


Audibility of small differences in amplitude over a wide frequency band (Low Q) is very noticeable and needs to be carefully addressed in the design on the speaker. On the other hand, audibility of small differences in amplitude over a narrow frequency band (High Q) is not very audible. The DSP in the LFR1100 already incorporates this High Q correction across the frequency band. Low Q correction is done in using the passive crossover network located inside the cabinet.

In response to all these inquiries we have created a DSP component that is able to do this High Q correction to the M80v3, VP180v3, VP160v3, and recently announced M100. The graphs below show the difference in the Listening Window response to an M80v3. Since the correction applied from the DSP is concentrated completely in the High Q group we would recommend this product for the discerning audiophile only as the audibility of this correction will be quite subtle.

The other feature on the DSP for the LFR1100 is the Boundary Switch, which is used to compensate for having your speakers very close to walls or large cabinets. Even though the M80s and M100s are not omnidirectional the proximity to boundaries still does have an effect in the lower frequencies. For that reason we have included a Boundary Switch on the DSP for these models also which should be experimented with in your room as the effect of the rear wall proximity will vary from room to room.

We will have these DSP Correction boxes available at the end of August. For anyone purchasing one of these to be used with their current M80v3 and VP180v3 we will need you to provide us with the serial numbers as there have been modifications to them along the way. We have created a spot to do this during the order process. The cost of the DSP box will be $950 for your two main speakers and $1,260 if you wish to add another DSP in the box for your VP180 or VP160. We will be doing a pre-release promotional price of $690 and $890 respectively from now until the end of July.

Broadband Digital Signal Processor

Uncorrected M80


Corrected M80

Would the difference be more pronounced on lower quality recordings, radio broadcasts etc?

The difference should remain the same no matter the type of recording you are playing.

I guess this is as good a thread as any to ask about a particular technical detail of the DSP box.

At what sampling rate does it operate internally? And how does it deal with any input signal that is at or above it's Nyquist limit?

I'm perfectly OK with a sampling rate in the 40 kHz range. Though, unless there's a steep decimation filter before the ADC, any Blu-rays foolishly encoded at 96 or 192 kHz may cause audible aliasing artifacts when resampled at that lower rate.

Hi ClubNeon,

As you are likely aware, there are three main components to the processing chain inside the DSP box: ADCs, DSP core, and DACs. Both the ADCs and DACs are Sigma-Delta, 24-bit, and operating at a 96kHz sampling rate. Inside the DSP core itself most of the processing instructions are double precision, 56-bit (28-bit × 28-bit multiplier with 56-bit accumulator). The DSP core is capable of performing 50 MIPS and we are utilizing a fraction of the available headroom. There is an active anti-aliasing filter at the input to the ADCs.

I hope that answers your questions!

Not only did that answer my question, it's the answer I wanted to hear.

If Chris is happy, then I'm happy.

Geeks.

So what this DSP box is doing is to convert the analog signal in digital (with the ADC chip), process it, and reconvert to analog (with the DAC chip) and give the corrected signal back to the amp?

P.S. So will it have a direct digital entry or only an analog entry? Would be nice to see the back of the box on this page: http://www.axiomaudio.com/broadband-digital-signal-processor#

The one for my LFRs only has analogue (though both RCA and XLR). I can't see how this would change in the new version. In fact the specs only mention the RCA and XLR inputs.

So let me see if I understand what Ian and Andrew are trying to explain: the DSP box essentially reduces some small differences in loudness between nearby frequencies as reproduced by the speakers. I guess this would be most noticable if you were playing music that included something like scales, or hitting each key on a piano sequentially. Even then the biggest difference (without DSP) should be something close to 5 db in the 60-90 Hz range which would be corrected to something less than 3 db by using the DSP.

This is a very subtle difference indeed.

Thank you cohesion. That is what I thought too (and effectively that is what is in the spec).

Andrew, if you're still watching this thread I thought of something that might be nice for people to see.

Since you have different DSP routines for the different speakers, can you make a "null" routine? One that just passes the input to the output?

Load that, and post a sweep showing how transparent the ADC-DSP-DAC signal chain is in the audible frequency range.

As for a direct digital input, the above should show how unnecessary it is, but also it would be difficult to make use of it. First you'd need something doing digital volume control, unless you like ALL REFERENCE LEVELS ALL THE TIME. Plus with HDMI and HDCP it's difficult to get a digital output for most media these days.

That brings another question: I wonder what would be the "upgrade" in taking a signal already converted (from digital to analog) using a multi thousands $ DAC (from Linn) that goes into another multiT$ Linn preamp to make it reconverted back from a $950 Axiom audio that does what an equalizer would do (if I understood correctly what the DSP is doing at the end)? Correct me if I am wrong but would you not say that the first graph which shows the reponse of the speakers depends on the quality of the signal going out of the preamp (and that is why you are correcting the signal). So DSP box will improve the sound if you have a lower quality source (DAC or other) and a lower quality pream than what the Axiom DSP is. I know money is not everything (your speakers are the proof of that) but still each time I tried to hook less expensive electronics on my Linn system that are supposed to improved the sound, it did the reverse (ie hugely downgrading the sound quality).

ClubNeon, I was asking about the direct digital input because there is still an audiophile market for good external DAC. So if the DSP can already convert digital signal to analog better than most low end CD player, streamer or whatever you are using as a music source, why not trying to get a share of this market? But I understand it could require some modifications in the circuitry.

P.S. HDMI and HDCP are for video contents. I was not talking about home theater but only stereo listening. Most hi-end audiophile product don't even have HDMI. Some mid end like Rotel or Arcam have it.

I think you're missing the point, oversimplifying, and on the verge of audiophile trolling.

First I do not know what audiophile trolling is (note that english is not my first language) but I suspect this is not a compliment. And second I don't find your anwser very usefull because it does not explain anything. So if the point of the DSP box is not to correct the signal before the amp sends it to the speakers then tell me what is it? Note that I understand the DSP box is technologically different than an analog equalizer and is achieving everythnig more precisely. But the goal is the same no? (in this case adjustment of the Q which determines the sharpness of the bandwidth and eventually to compensate for the room's acoustics. But I also guess the graphs here are from an anechoic room).

It's just that I think you'll find most people here would find a multi-thousand dollar DAC a complete waste of money.

Yeah, this DSP box probably has a DAC that measures better than a $10 CD player, but what can be measured, and what can be heard are two different things.

Any modern DAC will far outperform what the human ear can detect. An ADC/DAC pair will pass a band-limited signal without any audible change, and virtually no measurable change. The only place where they man be some noise creping into the system will be in the analog input/output, but that will likely be more than 90 dB below the signal, even further down when using balanced connectors.

Thanks for your answer ClubNeon. And I agree that what can be measured and what can be heard is very very different. That is why I have always said that the most important component in a sound system is the room! Regarding the waste of money, I understand what you are saying. But then that is until you've listen to some! And I am talking about A\B switch blind testing. My listening experience over the years showed me that the source is as important as the speakers. And frankly if I would be someone who judge the audio quality only by its price, would you think I would have bought Axiom speakers?

Charles, welcome. The DSP correction as used with the LFR1100s is designed to alter loudness, frequency and phase differences between the output of the front and back drivers. The purpose is to reduce the conflicts between front and back output which occur when those outputs are identical(as they are in some less sophisticated omni-directional designs).

As to some of the other points raised, both theory and personal experience indicate that there's no real audible(as distinguished from measurable)benefit from so-called "high end" DACs and other processors. A great thing about the current audio scene is the availability of electronic components which give us audibly flawless performance at quite low cost.

John is quite correct to point out that in an LFR speaker system the DSP performs some essential functions that would otherwise be infeasible to implement. In contrast, these functions are not all needed in non-LFR systems and the only function that is provided by the DSP in such case has only a very subtle effect. Given that the added cost of the DSP is a fairly substantial sum it may be justifiable only to those most comcerned with achieving the absolute best possible sound.

Hi Chris (and Charles65),

I can happily (and will do so later today) post some graphs of our DSP box in straight "pass-thru" mode. Honestly, there will be nothing to see in the amplitude response curve until you approach the Nyquist Frequency (f_s/2) where there will be a steep roll-off. Distortion and noise are also of a magnitude below that of most amplifier stages.

I do understand the concern anytime something is inserted within the analogue signal path, and the timing of your posts is excellent as we are going to be embarking on a controlled test in late August that will be of interest to many of you. Basically will are going to setup a reference front end system comprising a calibrated open reel deck on which we will be playing 2 track mix-down master tapes. For many this is considered the absolute ultimate in purely analogue source material. The signal from this deck will pass directly to a Bryston pre-amplifier, or through our DSP box running in "pass-thru" mode. Because our DSP is unity gain no level matching will be required, and all switching is performed with high quality, silver contact signal relays. Listeners will be able to switch between the two signals at will and can listen via loudspeakers, headphones, or their own supplied headphone systems. We will have a Bryston headphone amplifier for headphone listening. Ian will supply the specific details as we nail down firm dates and have procured some master tapes, but this session will run for a number of weeks and be open to ANYONE that would like to make the trip here to participate.

The question of digital inputs, as Chris correctly points out, is fraught with problems and only removes the ADC component from the equation. Simply trying to determine all of the required inputs and format standards results in a device that would be riddled with inputs and additional cost, not to mention ever-changing formats and HDCP requirements. In that sense it makes much more sense for us to stay within the analogue chain.

Thank you John (and Andrew and Chris also) for all your (DSP box) explanations. I understand better what it does. You are all much more knowledgeable than me in electronics. But I would have to disagree that above a certain limit all electronics are equal (if I understood well what you are suggesting). They might measure the same, as Celine Dion singing a note at 400Hz will measure the same as Aretha Franklin singing a note at 400Hz but you will agree that it won't sound the same to your ears! I do not know if my analogy is valid here. I guess what I am saying is that I do not think measurement is an absolute criteria. Are you saying that as long as it measures the same, your ear could not make a difference and the only difference would be in someone's head? As a scientific (not in electronics obviously! ), I am very open to this argument but frankly It is very far from corresponding to my lifelong empirical observations. Note that I am not saying that the price is an absolute criteria either and there is a lot of charlatanism in electronics. But having listened to a lot of DACs (streamers, processors, simple DACs) it is impossible for me to say they all sounded the same. It would be like saying all apple varieties taste the same because they are apples after all. It is just not what the chemical reactions in my brain is telling me.

Your analogy is invalid, just FYI. It is impossible for a human to produce a tone that lacks overtones/harmonics, which is what make people singing the same note sound different.

Interesting, thank you. So what you are saying is that there would be a measure that could explain the differences between what we hear. But this measure would not be the note or the fundamental tone at 440Hz. Would there be measures that could explain the differences between DACs chips? Because when I look into the specs of Wolfson DAC chips for example http://www.wolfsonmicro.com/products/dacs/ ,
it seems to me that the differences in SNR ratio for example are quite substantial and possibly audible (i.e not all DAC will sound the same since I think that 20-30 dB differences should be easily audible to the ear no?).

If the SNR was 20 or 30 dB you'd be able to hear that, but the difference between 98 and 128 is meaningless. Because you can't hear noise that is 98 dB below the signal in the first place. In a very, very quiet room, you may be able to hear noise that is 84 dB (14 bit) below the signal.

Oh, and it's not a "measure" of what's different between two voices (be they human or instrumental) producing a sound with a fundamental frequency at 440 Hz. They are two completely different waveforms. It's only the slowest part of the waveform that's cycling at 440 times a second. That's what gives any instrument its unique timbre.

But it is possible to digitally capture and accurately reproduce any of those sounds that exist below the Nyquist of the sampling system.

Nice work, Chris.

No, Charles; the point is that the measurements won't be precisely the same when instruments far more sensitive than our ears are used, but we can be unable to distinguish these small differences in controlled listening tests.

Thanks to you all for those explanations. Still it does not explain why I sometimes hear so much differences between DAC's (and processors in general). And I am not talking about subtle differences. It is true that between some DAC's the differences are very subtle that it would be possible that I imagine some (différences). But in many many cases, the difference is so in my face that you would have to be deaf or hypnothize to not hear it. For example, with my old Cambridge DAC magic (which I think is an excellent DAC for the price), I could hear very remotley Cecilia Bartoli taking her breath in (it was like in the background somewhere). Now with my Linn Majik, it is like she is breathing in my face. You clearly hear her respiration. It is impossible to not hear the huge difference. In some other recording, I could hear clearly the skweaking of the pedal drum with one DAC and not at all with the other one. So my deduction is that if I can hear (and everyone in the room also) this pedal skeaking it must exist in the recording. And if a source is not able to reveal that from the recording (giving all other Equipment the same), what else it is not revealing....

The best ways I've found to discover never heard before details on a recording is to either get a new piece of equipment, and listen intently, or to have someone point out minutia that they've heard when also doing intense listening.

So in a room of people you say, "wow, now I can hear the squeaking kick drum pedal," just as you first notice it, and everyone else will go, "oh, wow, I do too, that's amazing."

I can't say for sure what you've heard before, and the listening conditions. But this is why true double-blind testing is needed. Once you hear a new detail on one piece of equipment, you must be able to quickly switch to the other, and see if you can hear the same thing, and then back again. But you can't know which you're testing at any point.

Chris, I agree with all you said on the blind testing and did many times A\B switch blind testing in my life. But in the particular case I have in mind and that I mentionned, in one case you would hear it (the pedal skweak)) and the other case you would not at all. The A\B switch or double blind was not even necessary. Now is it a good thing. Not necessarily. I also remember another case with A\B switching where an old nakamichi CD player "beated" an very expensive processors (audio note). So I am absolutely not saying that more expensive is better. I am saying that not all sources are equal. And I also think that many people overlook the quality of the source. I also believe that your ear get use to a sound, so when you listen to something new, your brain is telling you that it is not as good.

Anyway, I would have like to buy the LFR1100 but I do not have the room (for now) to really enjoy them. So I bought the new M100 instead. Cannot wait to receive them and try them out!!! . I do not think I will buy the DSP box for now as I think it could be more usefull on the LFR1100 (from all your explanations) and I am not convince it could improve my listening experience with the room I have now (I would have to do an A\B switch test).

Charles, the digital to analog conversion process has no mechanism by which it could magically(even if the seller labeled a DAC "Magic" or "Majik")bring out previously unheard or almost inaudible material. That can happen when the loudness level is higher.

You'll undoubtedly enjoy your music through the M100s, and it's probably right that the DSP correction of the speaker(not the preceding electronics)which is at best extremely subtle, wouldn't be a wise purchase.

That's a good point. When I got my flagship DVD player that was "tuned" by Air Studios, and listened to its analog outputs vs. the digital connection to my receiver I was blown away in the difference in sound.

Turns out Air's tuning was mostly about running the RCA jacks at about -8 dBV, instead of the more usual -10 for single-ended, line level. I enabled the volume trim settings, and its unity position was actually correct. So with the volume control off (the default) it was running the outputs hot. Louder always sounds better.

The speed at which my receiver reacquires an HDMI sync is too slow for any quick flipping between inputs. With the extra Voltage disabled, and the 3 second delay between switches, I can no longer tell a difference.

John (and Chris)

I was not thinking that only the DAC chip per say might have caused the difference but also the rest of the circuitry. I made sure thought that the volume level was the same (because I also experienced what Chris described very well). What do you guys think about the quality of the power supply and its ability to keep the current constant (whatever the electronic component, amp , preamp, processors etc...)?

P.S. Well you don't have to answer if my questions are bugging you.