Does DSP belong in State of the Art Systems?

[Kal Rubinson]

There is no need to place a piece of gear in between the output of your DAC and the next piece of gear in the chain. Once the room is measured that digital measurement is altered with proper DSP/Correction software (Audiolense..etc) with a good PC that then is used to create correction filters at 24/192 or better. Those digital files are then applied to the source .wav, .flac files using convolution option in ROON or jRiver..etc. All the correction is done before the DAC even see's the signal

Agreed. To emphasize the phrase again, with digital sources, the DSP is applied directly at high resolution without any conversions. It is bit-perfect data processing and the output is suitable for conversion, amplification and transduction by the best equipment equipment in the world.

 

[Mark Seaton]

Agreed. To emphasize the phrase again, with digital sources, the DSP is applied directly at high resolution without any conversions. It is bit-perfect data processing and the output is suitable for conversion, amplification and transduction by the best equipment equipment in the world.

Quoted for emphasis. We can use what ever top quality DAC's we want. If someone uses a PC music server, there is no insertion loss. We can of course how that DSP tool is used, and uses range greatly. In exceptionally symmetrical and acoustically balanced rooms with well behaved speakers, less is more. Even in those rooms there can be room to fine tune and I'd argue that in the right hands such adjustments are far superior and precise than some of the fine tuning resistor adjustments I see on some speakers like the big Wilson's. The biggest difference is there's much more rope to get tangled up in vs 2-4 limited options to choose from.

DSP doesn't have to be full impulse response correction and attempts to fix acoustic problems. Today we have dramatically better tools than in the early days of a bench top RTA with LED strips which are available for less than the audio connectors and power receptacles many are using. They require knowledge, experience, and skill to get the most from them, but that's true of most any specialty tool.

Yes, there are some who get caught up in the advertised "automagic" capabilities of DSP. Anyone who well understands optimizing an audio system should know the best results come from getting the most from the physical setup and minimizing unwanted acoustic problems, especially at higher frequencies. What DSP or any signal EQ/processing can afford you is an added range to consider during the physical setup. A great example is when you face setup issues resulting in significant depressions/dips in the bass energy at the listener (say below 200Hz). In some cases there will be other placements which will fill in the worst valleys, but leave a peak or bump which might be objectionable in listening. With DSP available in the digital chain you can directly smooth that peak and gain back the detail and improve the balance that would have been present with the valley in the response. With the constant push for purity of signal, leaving most of the final voicing choices to the speaker and room interaction, most audiophiles greatly underestimate the audible significance of subtle shifts in broad spectral balance. Given the wide ranging sizes of our listening rooms and energy decay vs frequency, such adjustments can be invaluable if not overused. Many of the big 2-piece reference speakers with separate woofers are able to use the separate woofer and crossover adjustment to execute this bit of matching to the space, but because it's at the speaker level the end result adjustments get overlooked from a system perspective.

As I mentioned before and more than a few have echoed, the top point of focus with any digital adjustments should be if and how analog signals have to be made digital. The quality of the output DAC should be given the same consideration as in any state-of-the-art system. Signal levels are critical for optimal ADC, so putting this after a volume control needs to be considered carefully. The wider Voltage range we have to account for, the less optimized the conversion, and the more detail which can be lost to accounting for maximum levels. Source conversion to digital has the benefit that we have known signal ranges coming out of say a phono stage or R2R player. The only case where I find the post volume control ADC a worthwhile compromise is in the subwoofer range, but if given the option I would always prefer this done upstream.

 

[Mark Seaton]

it's not like you can put Humpty Dumpty back together again.

once the dsp gets ahold of it, the whole 'native' part is history. it's now, at best, the resolution/bit depth of dsp engine. if you did change it back to the original native format after dsp, the 'native' part is not relevant. it does not improve with another conversion. you are going backwards.....in truth to the source. the dsp 'action' might be a net gain depending on the system context. but there was a cost.

better to not need dsp.

again.....we are discussing 2 channel state of the art as a target. the little stuff matters.

Hi Mike,

As others have mentioned, it all depends on where in the chain and how such adjustments are made. If the signal was not otherwise going to be digital, that conversion has to be considered carefully.

I might be mistaken, but do I remember correctly that the amplifiers in your woofer towers have DSP on the input that the designer used for fine tuning of the speaker's response and blending with the full range speaker in your room?

 

[wil]

Quoted for emphasis. We can use what ever top quality DAC's we want. If someone uses a PC music server, there is no insertion loss. We can of course how that DSP tool is used, and uses range greatly. In exceptionally symmetrical and acoustically balanced rooms with well behaved speakers, less is more. Even in those rooms there can be room to fine tune and I'd argue that in the right hands such adjustments are far superior and precise than some of the fine tuning resistor adjustments I see on some speakers like the big Wilson's. The biggest difference is there's much more rope to get tangled up in vs 2-4 limited options to choose from.

DSP doesn't have to be full impulse response correction and attempts to fix acoustic problems. Today we have dramatically better tools than in the early days of a bench top RTA with LED strips which are available for less than the audio connectors and power receptacles many are using. They require knowledge, experience, and skill to get the most from them, but that's true of most any specialty tool.

Yes, there are some who get caught up in the advertised "automagic" capabilities of DSP. Anyone who well understands optimizing an audio system should know the best results come from getting the most from the physical setup and minimizing unwanted acoustic problems, especially at higher frequencies. What DSP or any signal EQ/processing can afford you is an added range to consider during the physical setup. A great example is when you face setup issues resulting in significant depressions/dips in the bass energy at the listener (say below 200Hz). In some cases there will be other placements which will fill in the worst valleys, but leave a peak or bump which might be objectionable in listening. With DSP available in the digital chain you can directly smooth that peak and gain back the detail and improve the balance that would have been present with the valley in the response. With the constant push for purity of signal, leaving most of the final voicing choices to the speaker and room interaction, most audiophiles greatly underestimate the audible significance of subtle shifts in broad spectral balance. Given the wide ranging sizes of our listening rooms and energy decay vs frequency, such adjustments can be invaluable if not overused. Many of the big 2-piece reference speakers with separate woofers are able to use the separate woofer and crossover adjustment to execute this bit of matching to the space, but because it's at the speaker level the end result adjustments get overlooked from a system perspective.

As I mentioned before and more than a few have echoed, the top point of focus with any digital adjustments should be if and how analog signals have to be made digital. The quality of the output DAC should be given the same consideration as in any state-of-the-art system. Signal levels are critical for optimal ADC, so putting this after a volume control needs to be considered carefully. The wider Voltage range we have to account for, the less optimized the conversion, and the more detail which can be lost to accounting for maximum levels. Source conversion to digital has the benefit that we have known signal ranges coming out of say a phono stage or R2R player. The only case where I find the post volume control ADC a worthwhile compromise is in the subwoofer range, but if given the option I would always prefer this done upstream.

Mark, if you have at your disposal both speaker level dsp with active subwoofers (after volume control) and computer level dsp (before volume control) would it make sense to blend the two? Or better to neutralize the speaker dsp and use only the computer dsp (like Audiolense or Accourate) instead?

 

[Bruce B]

Agreed. To emphasize the phrase again, with digital sources, the DSP is applied directly at high resolution without any conversions. .

I will have to disagree with this. DSP.... (plug-ins) work at a specified sampling rate. "Most" plug-ins work at 96/24.... if your source is 192/352.8/DSD, then you are down sampling! The same can be applied to hardware boxes as well.

 

[musicfirst1]

So, if I want to play with delay between the main speakers and the subs in my digital based system, what other options are there besides my MiniDSP 2x4?

 

[marty]

So, if I want to play with delay between the main speakers and the subs in my digital based system, what other options are there besides my MiniDSP 2x4?

You might want to try the simplest solution first. Is it possible for you to being the subs forward so that they are in the same plane as your mains? The need for time alignment is to get an arrival time at the listening position to be the same for the mains and sub presumably placed at some distance (2?,5? or even 8+? feet) behind the mains. DSP does this by retarding the signal propagation to the mains. This is what unfortunately may compromise the sonics to the mains when using even the best DSP instruments (and which may also occur, by the way, with any analog crossover ). If your room frequency response is good, try bringing the subs up to the mains. If you have the room, it's easy to try and may be a win/win scenario.

 

[wil]

I will have to disagree with this. DSP.... (plug-ins) work at a specified sampling rate. "Most" plug-ins work at 96/24.... if your source is 192/352.8/DSD, then you are down sampling! The same can be applied to hardware boxes as well.

Do you hear better sound at higher res than 96/24? Sincere question.

 

[Kal Rubinson]

I will have to disagree with this. DSP.... (plug-ins) work at a specified sampling rate. "Most" plug-ins work at 96/24.... if your source is 192/352.8/DSD, then you are down sampling! The same can be applied to hardware boxes as well.

That is certainly true. In general, I prefer the filtering/DSP tools in HQPlayer or Jriver. The only plug-in I am using is Dirac and there are plugins for all the sampling rates up to 192. For PCM (and DSD), it is always a challenge to decide whether the advantages of downsampling (or conversion to PCM) are more pleasing than playback at the higher resolutions. In general, they are.

 

[Bruce B]

Do you hear better sound at higher res than 96/24? Sincere question.

Yes..... but (excluding DSD)

Some DAC's and chip sets have a sweet spot where they sound the best. It's not as bad now as it used to be, but "generally"... most DAC's sound better either at 44/88/176/352/DSD or 48/96/192/384. I have never heard a DAC that can do both equally.

There are too many variables... what was the source rate, were there conversions, even/odd integer, etc...

To answer your question... . I prefer higher rates to listen to AND to work with.

 

[Alrainbow]

Definitely. I have played a lot with DSP over the years and even a cheap Behringer DEQ 2496 can be very transparent if one uses it only in the digital domain and account for the jitter both into the device and out of the device. I had two Monarchy Audio DIPs, one before and one after, the EQ and found that both helped the sound to the point that taking the Behringer out made now impact in sound quality.

Now, I use a mini-dsp digital xo/eq and find that it can sound very very good with external DACs (the one I bought doesn't even have DACs).

Thats some serious noise being inserted there , I’m sure DSP can make bad better , it would do the opposite to good or very good systems ...

 

[morricab]

Thats some serious noise being inserted there , I’m sure DSP can make bad better , it would do the opposite to good or very good systems ...

What noise are you referring to?

 

[wil]

The Comments section below this Archimago blog is really interesting. I learned a lot here about DSP.

http://archimago.blogspot.com/2019/10/2019-update-basic-acourate-dsp-room.html?m=1

 

[Mark Seaton]

Mark, if you have at your disposal both speaker level dsp with active subwoofers (after volume control) and computer level dsp (before volume control) would it make sense to blend the two? Or better to neutralize the speaker dsp and use only the computer dsp (like Audiolense or Accourate) instead?

Assuming you mean separate subwoofer outputs from the PC and not just L/R full range, then if state-of-the-art was the goal I would try to skip the ADC into the subwoofer amplifier. One exception would be if you could do a digital input to the subwoofer amplifier from the PC with a well executed digital volume control. To be fair, in most systems it's very possible to use the DSP in a subwoofer amplifier without any significant audible effects and much greater benefit due to the added adjustment and optimization. Many very impressive systems of members here do exactly this using subwoofers from Velodyne and JL Audio among many others.

The best answer to your question will depend on on the exact equipment and use.

 

[Atmasphere]

I would say that DBA for the range covered by the subs does the vast amount of what is needed in that range and DSP does the rest for the subs.

In saying that you would be incorrect. No amount of DSP can remove a standing wave! And its standing waves that cause problems with bass, especially below 80Hz or so. Standing waves can cause cancellation at certain frequencies so a loss of certain bass notes (along with others being emphasized). You can put all the power you want to into that cancellation and it will still cancel... In a nutshell, it doesn't work. DSP will work for removing peaks (since they are the result of superposition) but you can't do with nulls.

 

[sbnx]

Hello, Like a lot of audiophiles I have tried various forms of DSP going back to the Tact 2.0s in the early 2000's. I have also experimented extensively with various styles of subwoofer setup. When using the Tact I had a pretty nice setup (Eggleston Andra, Pass X600, Levinson 380s etc.) It wasn't SOTA for that time period but was up there. The problem was the room I was using was the family room and so there was very limited possibility for speaker positioning. Let's face it. DSP is not there to correct frequency response errors in your electronics or speakers (DEQX aside). It is there to correct errors caused by the room. Once you have achieved the best you can with the space you are in one simply asks "does the system sound better with or without DSP correction?". So does DSP belong in a SOTA system? That depends on how SOTA your room is.

Subwoofers offer a special case for DSP. One can DSP correct the signal going to the subs only without any "contamination" of the signal going to the mains. Also, the lower in frequency you go with the DSP correction the less audible it becomes. It is possible to correct both large peaks and nulls/dips in the frequency response with DSP and Subwoofers. Of course only below the crossover frequency you set for the subs. I offer the frequency response of my mains with and without DSP modified subs as evidence of this potential (with 1/24 octave smoothing; Sub crossover was at 40Hz). You can see the big, deep dip I get due to SBIR starting at 40Hz with the mains only. WIth subs I can fully fill that in. (Note that the dip below 25hz is there because I put it there as my fundamental room mode frequency is 23Hz.)

 

[DSkip]

Some things that come to mind for me:

1. I've never hear a PC-based system wow me. Perhaps I haven't heard the right one yet.
2. Any time I have played with DSP in my system, the soundstage simply collapses on itself. I do see benefits in certain systems, but at SOTA, I also believe a well balanced room is the key and not more processing, either prior or after conversion.

 

[Tim Link]

I agree dsp is not always desirable. It may have no place in an excellent room well matched to an excellent high end system. I am convinced that digital source material and whatever dsp went in to it's playback is technically superior to anything done analog, but that may not necessarily translate into a meaningfully more enjoyable listening experience. DSP can be very convenient and I think it can sound excellent.

 

[Kal Rubinson]

In saying that you would be incorrect. No amount of DSP can remove a standing wave! And its standing waves that cause problems with bass, especially below 80Hz or so. Standing waves can cause cancellation at certain frequencies so a loss of certain bass notes (along with others being emphasized). You can put all the power you want to into that cancellation and it will still cancel... In a nutshell, it doesn't work. DSP will work for removing peaks (since they are the result of superposition) but you can't do with nulls.

Please note that I said that DBA "does the vast amount of what is needed in that range" (not DSP) and, if done properly, will deal with both peaks and nulls. Below is the in-room low-frequency response of a single JL Audio f110v2 subwoofer (dotted red trace) vs three of the same subs (solid blue trace). There is some improvement in smoothness below 60Hz but more notable is the substantial mitigation of the nulls at 66Hz and 84Hz. See https://www.stereophile.com/content/jl-audio-fathom-f110v2-powered-subwoofer

 

[Atmasphere]

Please note that I said that DBA "does the vast amount of what is needed in that range" (not DSP) and, if done properly, will deal with both peaks and nulls.

My apologies- I read that the other 'way 'round. The peaks and valleys would be even smoother if 4 subs were used.

A nice side effect due to psycho-acoustics is that when you get the bass right, the mids and highs are immediately seeming to be faster and more transparent.