• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Dirac ART is now running on beta FW for Denon Xx800H AVRs!

I don't think it's wise to trust the words of companies. Plus I don't think they actually promised it.

I simply think that they stand to make more money by releasing it than by not releasing it.

Situation: You are company in low margin/mature market [not too many new features available] - your newest product is 2 years old. There is one interesting opportunity to significantly upgrade your product - in our case SW provided by 3rd Party.

Options:
1) give option for upgrade of existing product, which will significantly enhance usability of units out at the customers and prolong their usability. Your revenue stream is basically % of SW license revenue of 3rd Party provider. Your upcoming product will offer only very limited improvement and no real reason to upgrade.

2) introduce “new, better with unique feature” [which you can do at low cost, as you will probably re-use 90% of design of previous generation] let’s call it X900 - and you get flurry of attention, reviews, people talking about night and day improvements. There is significant incentive for X800, X700, X600 owners or even owners from competition to look at your new shiny product. Your revenue stream is - your product AND cut from SW license. Then 1 year later you can introduce FW update to older units and get % of license fees.

Actually 2) is what any sane company would do.
 
Situation: You are company in low margin/mature market [not too many new features available] - your newest product is 2 years old. There is one interesting opportunity to significantly upgrade your product - in our case SW provided by 3rd Party.

Options:
1) give option for upgrade of existing product, which will significantly enhance usability of units out at the customers and prolong their usability. Your revenue stream is basically % of SW license revenue of 3rd Party provider. Your upcoming product will offer only very limited improvement and no real reason to upgrade.

2) introduce “new, better with unique feature” [which you can do at low cost, as you will probably re-use 90% of design of previous generation] let’s call it X900 - and you get flurry of attention, reviews, people talking about night and day improvements. There is significant incentive for X800, X700, X600 owners or even owners from competition to look at your new shiny product. Your revenue stream is - your product AND cut from SW license. Then 1 year later you can introduce FW update to older units and get % of license fees.

Actually 2) is what any sane company would do.
My view on the situation for Denon/Marantz is somewhat different: You are company in low margin/mature market [not too many new features available] - your newest product is 2 years old. There is one interesting opportunity You promised to significantly upgrade your product - in our case SW provided by 3rd Party - almost 1 1/2 years ago. The last years have been turmoil. Your customers are unsettled and no longer trust you. The continued existence of the Denon/Marantz brands and the future development of the AV receiver sector are also uncertain.

Actual options:
1) give option for upgrade of existing product, which will significantly enhance usability of units out at the customers and prolong their usability. Your revenue stream is basically % of SW license revenue of 3rd Party provider. Your upcoming product will offer only very limited improvement and no real reason to upgrade.
2) introduce “new, better with unique feature” [which you can do at low cost, as you will probably re-use 90% of design of previous generation] let’s call it X900 - and you get flurry of attention, reviews, people talking about night and day improvements. There is significant incentive for X800, X700, X600 owners or even owners from competition to look at your new shiny product. Your revenue stream is - your product AND cut from SW license. Then 1 year later you can introduce FW update to older units and get % of license fees.
3) Do 1) and also introduce new products with more computational power expanding DiracART's capabilities. Doing so you won't break any promises and still sell numerous receivers to those who want the maximum feature set.
4) Screw your customers, introduce a complete new product line with DiracART, diminishing the residual value of the former generation products which don't get any further upgrades.

Actually 1) or 3) is what any sane company would do.
Maybe I am too pessimistic, but my guess is Denon/Marantz will choose option 4. In this case my hope is that a competitor will show up offering the same or even better performance.
 
Situation: You are company in low margin/mature market [not too many new features available] - your newest product is 2 years old. There is one interesting opportunity to significantly upgrade your product - in our case SW provided by 3rd Party.

Options:
1) give option for upgrade of existing product, which will significantly enhance usability of units out at the customers and prolong their usability. Your revenue stream is basically % of SW license revenue of 3rd Party provider. Your upcoming product will offer only very limited improvement and no real reason to upgrade.

2) introduce “new, better with unique feature” [which you can do at low cost, as you will probably re-use 90% of design of previous generation] let’s call it X900 - and you get flurry of attention, reviews, people talking about night and day improvements. There is significant incentive for X800, X700, X600 owners or even owners from competition to look at your new shiny product. Your revenue stream is - your product AND cut from SW license. Then 1 year later you can introduce FW update to older units and get % of license fees.

Actually 2) is what any sane company would do.
Increased parts costs right now. Cheaper to just add software and extract money from existing customers.
 
is 3) a real a real option?

I do not think there is something like ART Lite or ART Non Plus Ultra version - so what would be 6900 vs 6800 selling point, both having the same ART? 10W of extra power on each channel? I do not see any new video or audio feature that would make anybody buy 6900 vs 6800. X800 vs X700 was significant improvement - 4 subs, new UI, 8k, Dirac ready. But maybe I am missing soemthing.

Increased parts costs right now. Cheaper to just add software and extract money from existing customers.

All the D&M gear is heavily discounted in Europe since months, as it seems, that everybody that was interested to buy x800 has done it already Just by removing discount with new gen you make multiples of profits compared to licenses, and you can raise the price for new generation. And you keep your factories running and absorb fixed costs. In September it will be 3 years from x800. Anyway it is time to refresh the product line.

Playing the devil advocate here, not that I think I know anything, I don’t. Just saying what, I think, most rational option for profit driven company would be.
 
I do not think there is something like ART Lite or ART Non Plus Ultra version - so what would be 6900 vs 6800 selling point, both having the same ART?

The number of filter taps can differ from implementation to implementation. That’s how the Node Icon can apply Dirac in high resolution.

"At the StormAudio Dirac tech talk at MWAVE 2024, Matthew Trinklein mentioned the filter tap count on StormAudio allows finer filter resolution than competing products. The amount of filter taps on StormAudio is 3,076. The next product down in the count has 1,554, and most products have 1,100." Source

Trinnov has up to 130,000 taps per channel. Dirac on PC allows 65,536 taps maximum per channel.

When MiniDSP discusses the filter resolution, that is all channels. So a four channel product with 4096 taps has 1024 taps per channel.

My understanding is that a lot of this is unnecessary since you only need ultra high resolution in the lower frequencies and you can use regular IIR filters at low frequencies with good results historically. My understanding is that ART and Waveforming really push the precision requirements in the bass to the next level.

@TimoJ , can you comment on the HTP-1 beta with Dirac ART? My understanding is that it has fewer taps than the Storm Audio for Dirac ART.
 
@TimoJ , can you comment on the HTP-1 beta with Dirac ART? My understanding is that it has fewer taps than the Storm Audio for Dirac ART.
Sorry, I don't know technical details. But since StormAudio has 2 more DSP chips, I think it's safe to say that it has more taps. Then again, why does StormAudio have ART support channel limit with the Core 16 while the HTP-1 hasn't..?
 
The number of filter taps can differ from implementation to implementation. That’s how the Node Icon can apply Dirac in high resolution.

"At the StormAudio Dirac tech talk at MWAVE 2024, Matthew Trinklein mentioned the filter tap count on StormAudio allows finer filter resolution than competing products. The amount of filter taps on StormAudio is 3,076. The next product down in the count has 1,554, and most products have 1,100." Source

Trinnov has up to 130,000 taps per channel. Dirac on PC allows 65,536 taps maximum per channel.

When MiniDSP discusses the filter resolution, that is all channels. So a four channel product with 4096 taps has 1024 taps per channel.

My understanding is that a lot of this is unnecessary since you only need ultra high resolution in the lower frequencies and you can use regular IIR filters at low frequencies with good results historically. My understanding is that ART and Waveforming really push the precision requirements in the bass to the next level.

@TimoJ , can you comment on the HTP-1 beta with Dirac ART? My understanding is that it has fewer taps than the Storm Audio for Dirac ART.
From memory, Audyssey XT32 is specced at 512 taps? (total?)

DSP's from 2008 or thereabouts, had Audyssey XT32 capability....

Current (mass market) DSP's should be able to handle around 4x that without too much trouble... so say 2048 taps, that still leaves a substantial deficit compared to StormAudio's current processors!

Which will naturally bring us back to discussions centered around how much is enough, and what is needed under various circumstances

Mainstreaming ART, will immediately move the focus to DSP processing power, rather than "mere" functionality..... (the functionality having become a boring passe topic, now that everyone can basically do all the required functions!)
 
Too many taps will mess with latency for gaming and video. They can hold it for the video sync but point for gaming it would be bad
 
Too many taps will mess with latency for gaming and video. They can hold it for the video sync but point for gaming it would be bad
For gaming there will always be another balancing act between performance (latency) and level of EQ (# Taps)...

Hence I feel sure that once ART releases for mass market brands, it will kick off a DSP performance race...

Which may have some consequences in terms of AVR heat management and longevity/engineering design... there will be a risk of revisiting the nasties of the 2008 to 2015 era - where heat was a major bugbear.

DSP's will be pushed to higher clock speeds for performance, and designs will likely move from dual processor to quad processor DSP's (or more).

Inspecting the heatsinking of the DSP chipsets will become an important evaluation step once again.
 
Inspecting the heatsinking of the DSP chipsets will become an important evaluation step once again.
I strongly doubt that. Sound is a solved problem as far as processing power/efficiency are concerned, IMO. Current cutting edge is in video processing and rendering. That's were heat dissipation still matters.
 
Last edited:
The number of filter taps can differ from implementation to implementation. That’s how the Node Icon can apply Dirac in high resolution.

"At the StormAudio Dirac tech talk at MWAVE 2024, Matthew Trinklein mentioned the filter tap count on StormAudio allows finer filter resolution than competing products. The amount of filter taps on StormAudio is 3,076. The next product down in the count has 1,554, and most products have 1,100." Source

Trinnov has up to 130,000 taps per channel. Dirac on PC allows 65,536 taps maximum per channel.

When MiniDSP discusses the filter resolution, that is all channels. So a four channel product with 4096 taps has 1024 taps per channel.

My understanding is that a lot of this is unnecessary since you only need ultra high resolution in the lower frequencies and you can use regular IIR filters at low frequencies with good results historically. My understanding is that ART and Waveforming really push the precision requirements in the bass to the next level.

@TimoJ , can you comment on the HTP-1 beta with Dirac ART? My understanding is that it has fewer taps than the Storm Audio for Dirac ART.
Do you think that downsampling content above 88.2/96 khz to 88.2/96 khz or even to 44.1/48 khz would allow Dirac to perform at its best? (Even when using the more capable Dirac on PC?)
 
Do you think that downsampling content above 88.2/96 khz to 88.2/96 khz or even to 44.1/48 khz would allow Dirac to perform at its best? (Even when using the more capable Dirac on PC?)

Dirac already downsamples to 41-48 kHz in virtually all implementations. It’s rarer to find the high resolution ones, with the PC/Mac, NAD/Bluesound, Theta, and older generation Arcam being the ones that I can think of.

Like many things, more is better but not always. If you have enough filter taps to achieve your goal, you are set. Too many taps or too long of a filter and you may run into pre-ringing artifacts that are audible. Not enough filter taps means less precise correction, although the human ear is not as sensitive to frequency response irregularities once you are in multichannel.
 
In my case, I am using Dirac on PC, which doesn't downsample anything. But I guess those 65,536 taps per channel you mention are ok even for 192 khz tracks?
 
From memory, Audyssey XT32 is specced at 512 taps? (total?)

DSP's from 2008 or thereabouts, had Audyssey XT32 capability....

Current (mass market) DSP's should be able to handle around 4x that without too much trouble... so say 2048 taps, that still leaves a substantial deficit compared to StormAudio's current processors!

Which will naturally bring us back to discussions centered around how much is enough, and what is needed under various circumstances

Mainstreaming ART, will immediately move the focus to DSP processing power, rather than "mere" functionality..... (the functionality having become a boring passe topic, now that everyone can basically do all the required functions!)
There's a lot of misinformation about the number of FIR filter taps and how they function in AV receivers. Audyssey XT32 sends 1024 float values (filter taps) to modern receivers, but only uses 1003 for speakers and just 687 for subwoofers - the rest are simply zeros. Dirac DLBC sends 2048 floats to a different DSP chip in the same receivers, though I’m not sure how many of those are zeros, as I haven’t used Dirac extensively.

If these filters operated at 48kHz, neither would be suitable for effective bass correction. For example, Audyssey’s 687 taps for subs would offer a frequency resolution of around 70Hz, meaning they couldn’t accurately correct anything below that, or would allow for just one peaking filter say between 70Hz and 140Hz. However, both systems use well established multirate decimation techniques: they break filters into frequency bands, interpolate and process each at lower sampling rates (down to about 750Hz in the bass region). Receivers upsample them to 32kHz, 44.1kHz, or 48kHz depending on the source. This approach allows for precise convolution where it's most needed - without requiring thousands of taps.

The technique is highly efficient, and increasing the number of filter taps likely wouldn’t improve performance. Audyssey doesn’t even utilize all the available 1024 taps as I mentioned above. It’s also unlikely that DSPs will increase their filter capacity soon, since AV receivers must also buffer 8K video, and the latency required for more FIR taps adds to the cost and complexity.

Reference: Multirate Signal Processing - Chapter 9 (Oliver Hinton, Newcastle University)
 
Denon/Marantz has been sold to Harman. Hopefully this will clear the air and allow the company to operate normally.
 
Denon/Marantz has been sold to Harman. Hopefully this will clear the air and allow the company to operate normally.
This explains the hold-up on D&M products getting A.R.T as it's being sold at the end of 2025. So looks like our timeframe will be shifted again. I highly doubt they would add it to this current lineup. More than likely it will be comming out with the New Models then after some time may be available for download on this Gen of current products that support it.
 

Attachments

  • Screenshot_20250517_151621_Samsung Internet.jpg
    Screenshot_20250517_151621_Samsung Internet.jpg
    166.9 KB · Views: 55
  • Screenshot_20250517_151657_Samsung Internet.jpg
    Screenshot_20250517_151657_Samsung Internet.jpg
    200.1 KB · Views: 57
D&M along with Dirac committed to a 3Q 2025 release for ART
I am not stressing out any more. It will come when ready.

Dirac and D&M have less than stellar record when it comes to ART implementation and many posts were written on many forums over the years of anticipation. As I understand it, D&M would not want to support buggy ART, and Dirac is used to royal treatment by Storm that is supporting it fully - but in a completely different price range. So ART needs to be relatively uncomplicated for a brand like D&M to push it to their $1K+ AVRs.

It is encouraging to see that at least some other brands are announcing ART support. Pity it is only 16 channel processor, so 12+4. And smaller brand that is probably ready to support ART for the users.

 
Back
Top Bottom