Hyperion APR-16 AV Processor Review

[peng]

As @Fidji noted, they should have gone with 1 channel more just for a good measure. 16 channels continues to be an odd channel count despite the number.

I have to wonder if it is a cost efficiency thing, the fact that AVPs that use the 8 channel dac ICs, to add 1 more channel, it would mean adding another 8 channel dac IC, in order to have the same DAC IC for all channels. Easier for the AV10 and AV20 because those use 2 ch DAC ICs.

 

[Oddball]

I have to wonder if it is a cost efficiency thing, the fact that AVPs that use the 8 channel dac ICs, to add 1 more channel, it would mean adding another 8 channel dac IC, in order to have the same DAC IC for all channels. Easier for the AV10 and AV20 because those use 2 ch DAC ICs.

Well their choice. 16 continues to be an odd number as far as HT.

 

[welwynnick]

They have offered the DPR16 16x.

Now that's an offer that's too good to be true. The DPR-16 is probably the answer to all my questions.

It's a digital to digital converter, so the scope of testing would probably be similar to the Arvus H2-4D and the Storm ISP Evo.

What many of us would want to know is whether the DSP is limited to 48kHz like almost everything else, or can it process 96kHz or 192kHz?

 

[PGAMiami]

Not just them but key engineering team which jumped on this to fix the problem. I am sure they had other things to do.

Importantly, they did not do the usual complaining of why I had not run the review by them first. They saw a problem, and immediately went to work to replicate and fix it.

Very, very positive interaction. They even offered another test unit if I had to return the review unit back.

The “total package” does seem very compelling. Jury is out still on how it will perform compared to the Trinnov and Storm, but there is good reason to think the APR-16 will stack up nicely and at a more compelling price. It’s more expensive than a Tide16, but the street price difference can be less than people may think. There is something to say with having a product that has dealers and that is manufactured and engineered here in the US by such a service focused team. Left yet to discover are the little quirks of how it operates day to day, with hdmi, CEC, reliability, etc. That’s one area where Denon Marantz seems to have a very solid footing.

It will also be interesting to see how Dirac evolves over time on the APR-16 and its competitors. Is the algorithm on the APR-16 the same as that on a 16 channel Storm? Are there any future upgrades to Dirac that could run on the more powerful DSP in an APR-16 that may not run on a Marantz? Given how transparent the hardware is, the bigger questions I have are on the DSP capabilities of the unit.

 

[darth]

It is a pretty expensive product just to borrow for that reason. They have offered the DPR16 16x. It is a neat looking processor but seeing how it is all digital, I am not sure what I would be testing.

You would be making lurkers of this thread happy.

 

[Sam Ash]

Since the DPR-16 is digital, it would be interesting to know if there is a possibility to expand the number of channels.

 

[welwynnick]

Since the DPR-16 is digital, it would be interesting to know if there is a possibility to expand the number of channels.

Not over AES3, but maybe over Dante.

 

[prerich]

I have the AudioControl X9 and I have 4 subwoofers in my room. One issue that the X9 had when using Dirac Live with bass management is that if I plugged in a subwoofer to the XLR input that is labeled subwoofer, the unit did not recognize subwoofers that are placed on channels 14, 15 and 16. The only workaround was to not use the subwoofer connection and place two subs on 14 and 15 and to connect with a Y cable the other two subwoofers to channel 16. I have been waiting for the Hyperion 16 to come out but I can that the channels are the same and I do not want to get into the same situation. Also, I am concerned about the low frequency issue that a beta firmware update (?) has fixed....hmmm. What I appreciate about the Anthem and Marantz is that they have 4 separate subwoofer inputs.

I have the AudioControl Maestro x9s, and as per the manual if you use 4 subs, you should use channels 13,14,15, and 16. That's the way I have mine set up as I've added my ACI Titans back into the mix and have them in a seperate group from my HSU VTF-3 mk5hp's. It works perfect that way.

 

[Oddball]

The “total package” does seem very compelling. Jury is out still on how it will perform compared to the Trinnov and Storm, but there is good reason to think the APR-16 will stack up nicely and at a more compelling price. It’s more expensive than a Tide16, but the street price difference can be less than people may think. There is something to say with having a product that has dealers and that is manufactured and engineered here in the US by such a service focused team. Left yet to discover are the little quirks of how it operates day to day, with hdmi, CEC, reliability, etc. That’s one area where Denon Marantz seems to have a very solid footing.

It will also be interesting to see how Dirac evolves over time on the APR-16 and its competitors. Is the algorithm on the APR-16 the same as that on a 16 channel Storm? Are there any future upgrades to Dirac that could run on the more powerful DSP in an APR-16 that may not run on a Marantz? Given how transparent the hardware is, the bigger questions I have are on the DSP capabilities of the unit.

Very well put together. Given how fast Hyperion team fixed the issue discovered in review, I would hope that this AVP will pur as a cat in no time.

Not sure why are you mentioning Storm? There is nothing really special about Storm except the channel count. If you look at HTP-1, it actually out-gunns Storm 16 in features and apparently they measure similarly on SINAD bench.

I am not aware that Dirac ART currently has different algorithms. All certified gear should run the same algo. There are differences in number of cross support filters as D&M decided to implement less of them than needed to support all channels by all channels for their higher channel count models. So the difference is not in how it works, but what are limitations on cross-support. Not defending D&M, but for anyone familiar with ART, having anything support everything is questionable. There should be some rationale for assigning support groups. For example, some slice it by L and R side, some slice it by front/middle/back side.

 

[Kal Rubinson]

There should be some rationale for assigning support groups. For example, some slice it by L and R side, some slice it by front/middle/back side.

Those may seem logical but is there any way to prioritize them other than cut-and-try?

 

[Oddball]

Those may seem logical but is there any way to prioritize them other than cut-and-try?

Trial and error still the best science here.

I personally prefer the second approach, but perhaps also because L and R sides of my room are not acoustically identical (R open to other spaces). So LR supported by wides. FW supported by LR and surrounds. Surrounds supported by FW and SB. SB supported by surrounds. Subs support everything. Pretty graphs and sound very good. Kind of blending the support from front to back. With more support, I could potentially tighten the spreads a bit in 100-150hz range, but then I can't really tell the difference as spreads are already quite good.

 

[PGAMiami]

I don’t believe there is any confirmation the algo is the same and certainly not that it will stay the same. Why would Dirac lock themselves to the limited DSP capabilities of d&m?

 

[Oddball]

I don’t believe there is any confirmation the algo is the same and certainly not that it will stay the same. Why would Dirac lock themselves to the limited DSP capabilities of d&m?

They are selling the product that is called Dirac ART. They are not disclosing that there are different tiers of Dirac or different algorithms. They are acknowledging different implementation in cross filter support but still certifying D&M products. If indeed there was inferior algorithm running (not cross support filters) on D&M than on Storm, that would be an easy win for the class-suit against Dirac due to lack of due disclosure. I don't think they are that stupid, but one never knows.

I am certainly not saying that Dirac should stop developing or that it should do so due to D&M. D&M did not disclose if their implementation was due to hardware limitations or their approach that simpler is better. Given how much HTP-1 can do (that is older design), we can just make uneducated guess that D&M just limited the cross support filters because they did not want to confuse the consumers and face them with too many choices.

I am pretty sure that ART algo will not change much in the foreseeable future as it is really good as is. They need to work on 20hz barrier and on timing alignment with Dirac Live, but these are probably not hugely demanding changes. Dirac Live is really old product by now and honestly not that superior to other auto solutions. So IMO Dirac needs to find a new spice to revamp Dirac Live. There is this notion that one should EQ only up to room transition frequency, but that does not take into account various multi-channel systems that need EQ to the tilt to equalize response of different speakers in the system.

 

[azzy_mazzy]

From what i have seen on Matthew Poes videos the only products getting custom algorithms are the cars which are big projects and are done once. doesn’t make sense for home systems and complicates the work load.

 

[EWL5]

They are selling the product that is called Dirac ART. They are not disclosing that there are different tiers of Dirac or different algorithms. They are acknowledging different implementation in cross filter support but still certifying D&M products. If indeed there was inferior algorithm running (not cross support filters) on D&M than on Storm, that would be an easy win for the class-suit against Dirac due to lack of due disclosure. I don't think they are that stupid, but one never knows.

I am certainly not saying that Dirac should stop developing or that it should do so due to D&M. D&M did not disclose if their implementation was due to hardware limitations or their approach that simpler is better. Given how much HTP-1 can do (that is older design), we can just make uneducated guess that D&M just limited the cross support filters because they did not want to confuse the consumers and face them with too many choices.

I am pretty sure that ART algo will not change much in the foreseeable future as it is really good as is. They need to work on 20hz barrier and on timing alignment with Dirac Live, but these are probably not hugely demanding changes. Dirac Live is really old product by now and honestly not that superior to other auto solutions. So IMO Dirac needs to find a new spice to revamp Dirac Live. There is this notion that one should EQ only up to room transition frequency, but that does not take into account various multi-channel systems that need EQ to the tilt to equalize response of different speakers in the system.

My gut feeling is that the ART algorithms are going to be the same across all processors. What sets each brand/model apart are the MIPS possible from the onboard DSP and the cross terms will scale accordingly. It is truly an arms race and I'm sure Onkyo has to discard decades old DSP tech in order to compete in 2027.

 

[Oddball]

As noted MIPS possible could be just limited with firmware, not DSP.

Let's take a practical example of my system. 9.4.4 Marantz AV10. I run out of cross support filters if I want to support anything with everything. But then I get awesome results with the approach that I described a few posts above. Now, if I had unlimited filters with this channel count, like Storm would, then I can't see any practical benefit to it. On the contrary, I would have wasted 100 more hours trying to figure out if more support is better. Well in my system it is not. Perhaps some other system could benefit from more cross support. Dirac's interviews claim so, but that is not really what happens in reality. Tightening the tight spreads really amounts to nothing audible.

In general, the recipe for building great multi-channel ART system is 4 subs and then speakers that can take it from subs at 80-100hz range with sufficient SPL for your needs. 9.4.4 is often thought of the starting point for full resolution syste, which in this case would need to be scaled down to 9.3.4. With a bigger sub on the back wall and two up front this could likely be done without any audible penalty.

 

[PGAMiami]

You all are making many assumptions. Dirac has shown a pattern of continuous improvement. I would expect the algorithms will continue to evolve. They now have three products each significant improvement over the prior. Why not 4 or 5? Certainly they can launch higher resolution versions of the filters, more like what Audiolense is able to do on a PC. Why would Audio Control and miniDSP bother to use more capable processors?

 

[amirm]

Why would Audio Control and miniDSP bother to use more capable processors?

If you are porting (implementing the code) yourself, the more powerful the processor (and the more memory), the easier the job. With ultra low cost hardware, this becomes a heroic job that few are capable of doing. Hence the reason major AV companies use the off-the-shelf solution.

It is also possible that they don't use all the resources and can indeed add more functionality. Kind of hard to say.

My guess is that newer designs are able to use newer CPU/DSPs that did not exist when the original DIRAC implementations were rolled out.

 

[Oddball]

You all are making many assumptions. Dirac has shown a pattern of continuous improvement. I would expect the algorithms will continue to evolve. They now have three products each significant improvement over the prior. Why not 4 or 5? Certainly they can launch higher resolution versions of the filters, more like what Audiolense is able to do on a PC. Why would Audio Control and miniDSP bother to use more capable processors?

I am making assumptions based on my view - same as you are based on your view. Dirac has hands full selling what they have, and looks like they are making a pretty penny. There is no product that offers significant improvement to their initial offering of ART or Dirac Live. Where is Dirac going is what they know, but has not been disclosed.

Storm recently announced launch of new products, so if Dirac is planning new releases I would not be surprised if they would develop them with their most trusted partner. As far as I understand, there are no details as to what new Storm products will be or how they will run. Storm for sure could use refreshment appart for just stacking up channels.

Have no idea why Audio Control or Mini are using more capable processors, but it is certainly a good thing. Will they be able to run future Dirac releases, well that's another story.

Without any data to back it up, one could expect that Dirac revenues from D&M are a major part of their AVR/AVP market.

 

[PGAMiami]

If you are porting (implementing the code) yourself, the more powerful the processor (and the more memory), the easier the job. With ultra low cost hardware, this becomes a heroic job that few are capable of doing. Hence the reason major AV companies use the off-the-shelf solution.

It is also possible that they don't use all the resources and can indeed add more functionality. Kind of hard to say.

My guess is that newer designs are able to use newer CPU/DSPs that did not exist when the original DIRAC implementations were rolled out.

As I understand the Dirac ART algorithm may be the similar or even the same, but the number of taps and processing depth for each of the convolution filters is determined by the manufacturer. The DSP in a Tide16 and APR16 will surely allow for higher resolution filters than what the DSP in an AV10 and 3800H allow. Even if the number of filters is the same as it would be in a system with less speakers. It would be very interesting to get details either from Dirac or the manufacturers on the implementation of Dirac on their hardware.