Hyperion APR-16 AV Processor Review

[Labjr]

You didn't say this but to be clear, the latency cannot be dealt with using faster processors. You still have to wait for samples to arrive to process them. Therefore, the longer the filter, the longer the latency.

What you can do of course process more channels, or more accurate filtering, etc.

Delay the video.

 

[PGAMiami]

You didn't say this but to be clear, the latency cannot be dealt with using faster processors. You still have to wait for samples to arrive to process them. Therefore, the longer the filter, the longer the latency.

What you can do of course process more channels, or more accurate filtering, etc.

Agreed. In essence the DSP has to “hear” the note before it plays so it can start correcting it prior to it starting. I believe with J River you can delay the video so the latency of the audio remains in sync. When Mitch Barnett set up my filters he had one set with high latency just for music that ran on Roon for 2.0 stereo and another for video that ran on a PC. The music one sounded quite good from the midrange on up. But I believe I prefer the bass on ART with four subs and overall ART was better

 

[PGAMiami]

Amir I would bet dollars to donuts that there are many bugs probably worse than what you found when one engages Dirac ART. Not to mention that we have no idea really how well the math is being executed. Can you think of some basic tests you could run on the DSP other than measuring the actual sound in the room?

 

[amirm]

Delay the video.

Which costs RAM and adds to delay that is already in the display. Further, no one wants to hit play and wait seconds for it to show video/play audio.

 

[amirm]

Amir I would bet dollars to donuts that there are many bugs probably worse than what you found when one engages Dirac ART. Not to mention that we have no idea really how well the math is being executed. Can you think of some basic tests you could run on the DSP other than measuring the actual sound in the room?

Do you mean DSP or DIRAC? DIRAC is a long process as you know and once run, it messes big time with the response of each channel. So I can't think of a reasonable way for me to test it.

If you mean DSP such as setting up high pass filters and such, I can do that. The problem is, testing AVR/AVPs is already exceedingly long and tiring. Hence the reason it took a while to test yours. Adding more tests such makes a bad situation worse.

All of this said, I will think about it.

 

[azzy_mazzy]

high end TVs are 60ms+ in non gaming modes, highest i seen is 300ms!

 

[TimoJ]

Do you mean DSP or DIRAC? DIRAC is a long process as you know and once run, it messes big time with the response of each channel. So I can't think of a reasonable way for me to test it.

You just need a mixer and a USB sound card (or a mixer with a USB output) to create a "fake" room and microphone. For example, use an 8-channel mixer and configure the AVP for 7.1. Connect the AVP outputs to the mixer, route the mix output to the sound card, and configure the Dirac app to use it as the microphone.

Run 1-3 (3 for ART) Dirac calibrations through the mixer, save the results, and measure how the SINAD and other parameters changed.

 

[PGAMiami]

I tested with REW the total system response of the Audiolense convolution filters on Roon by measuring the room response with and without the filters. You could then see whether you were hitting the target curve, distortion, etc. I assume we may be able to do the same with Dirac? Of course this is measuring the whole chain, speakers, amps, room … and is dependent on the calibrated mic.

 

[amirm]

You just need a mixer and a USB sound card (or a mixer with a USB output) to create a "fake" room and microphone. For example, use an 8-channel mixer and configure the AVP for 7.1. Connect the AVP outputs to the mixer, route the mix output to the sound card, and configure the Dirac app to use it as the microphone.

Run 1-3 (3 for ART) Dirac calibrations through the mixer, save the results, and measure how the SINAD and other parameters changed.

Say what? How is this the answer to what I post regarding the work it takes to run Dirac? And no, you can't just look at how SINAD changes. SINAD will change based on EQ without anything causing performance change as it relies on noise bandwidth product.

 

[TimoJ]

Say what? How is this the answer to what I post regarding the work it takes to run Dirac? And no, you can't just look at how SINAD changes. SINAD will change based on EQ without anything causing performance change as it relies on noise bandwidth product.

Of course, running Dirac requires some extra steps. If you're not willing to do that, then I guess we'll never know how Dirac affects the performance measurements.

 

[PGAMiami]

Of course, running Dirac requires some extra steps. If you're not willing to do that, then I guess we'll never know how Dirac affects the performance measurements.

You just need a mixer and a USB sound card (or a mixer with a USB output) to create a "fake" room and microphone. For example, use an 8-channel mixer and configure the AVP for 7.1. Connect the AVP outputs to the mixer, route the mix output to the sound card, and configure the Dirac app to use it as the microphone.

Run 1-3 (3 for ART) Dirac calibrations through the mixer, save the results, and measure how the SINAD and other parameters changed.

I don’t think this would tell you much. Testing DSP is one thing. We can load a filter on the APR-16 and then measure how well it’s performing, testing Dirac is another thing altogether.

For starters it’s only going to be as good as the measurements. That depends on the operator, mic, room …Maybe a particular location is useful for some of your speakers but doesn’t have line of sight to other speakers …. Then we have to trust Dirac does the right calculation on its very complex filter design that I understand is a combination of biquad PEQ for the lower frequencies (to avoid latency and not run out of taps) and FIR higher up, plus judiciously applied delays. And that Dirac also gets right any absorption it plans to do. Not to mention that Dirac is assuming your speakers have sufficient dynamic range and low distortion (otherwise the adsorption and support will not be as Dirac expected).

And all this involves lots of math. Knowing how many filters Dirac is using is just part of the story. We also need to to know how precise is the filter calculation and how precise is the implementation. So until Dirac is more transparent on all this, best we can do is run some before and after REW measurements to see how well this is working.

One thing I can say is multiple people thought my room had never sounded better for 2 channel stereo (using multiple speakers supporting the bass) and movies, than with Dirac on a 4800h as a prepro.

My APR-16 had to go back to Audio Control. Seems the firmware may have gotten messed up. Good news is they have been extremely responsive, logging into my unit remotely to diagnose it and sending an overnight fedex label asap.

 

[TimoJ]

testing Dirac is another thing altogether.

For starters it’s only going to be as good as the measurements. That depends on the operator, mic, room …

That's why I suggested using a mixer. With that setup, Dirac measures only the line-level signal path through the mixer, eliminating the room, microphone, and operator from the equation. The goal isn't to evaluate how well Dirac corrects room acoustics, but simply to determine whether enabling it degrades the signal quality.

 

[amirm]

That's why I suggested using a mixer. With that setup, Dirac measures only the line-level signal path through the mixer, eliminating the room, microphone, and operator from the equation. The goal isn't to evaluate how well Dirac corrects room acoustics, but simply to determine whether enabling it degrades the signal quality.

Dirac requires differing measurements as it instructs you to move the microphone. If you keep giving it the identical input, then it may error out.

 

[TimoJ]

Dirac requires differing measurements as it instructs you to move the microphone. If you keep giving it the identical input, then it may error out.

I don't think it will error out. It doesn't do it with microphone in one place (but mic signal is never exactly the same).
But if it does, just perform a single Dirac measurement.

 

[BJL]

That's why I suggested using a mixer. With that setup, Dirac measures only the line-level signal path through the mixer, eliminating the room, microphone, and operator from the equation. The goal isn't to evaluate how well Dirac corrects room acoustics, but simply to determine whether enabling it degrades the signal quality.

This strikes me as completely meaningless. What you would be measuring is the mixer electronics and whatever it is being used to capture the audio. You won't be measuring Dirac unless you measure the audio post filters. None of what you suggest would provide insight into anything about Dirac in actual use, which is the audio processed through filters. Activating Dirac without filters will only measure that stage of the component, if anything, if you could possibly eliminate the contribution of the mixer etc.

The measurements (without processing) of the electronics typically done here is far more useful because they provide an objective benchmark for quality of the component.

Attempting to measure supposed degradation from Dirac is especially absurd in the case of ART, where you have a default set up, but then one can (and probably should) go back and customize the filters for one's own preferences. It is inherently subjective as the end result is the result that pleases the individual listener. What we care about with ART is what we hear. What we care about with the originating component is whether it passes the signal without distortion, whatever that signal may be, processed or not.

 

[TimoJ]

What you would be measuring is the mixer electronics and whatever it is being used to capture the audio.

The mixer is used only during the calibration. Why would it affect measurements taken afterward using only the AVP's HDMI input and line outputs? Those measurements would be identical to the standard ASR measurements, except that Dirac processing would be active using the filters created during the calibration. Is this really too complicated to understand?

 

[peng]

The mixer is used only during the calibration. Why would it affect measurements taken afterward using only the AVP's HDMI input and line outputs? Those measurements would be identical to the standard ASR measurements, except that Dirac processing would be active using the filters created during the calibration. Is this really too complicated to understand?

If I understand your point, it is about whether room correction software like DL would add distortions and/or noise. Never hurt to measure the effects if there are methodologies to do it accurately enough, but just curious, do you think chance is reasonably good that distortions/noise would increase enough for it to be audibly discernible, to people with no, to not much hearing loss? I have used manual EQ, Audyssey, AARCG, DLBC for a long time and never felt any of those RC/REQ increase distortions/noise, but I don't listen anywhere close to reference level.

Regardless, if it can be done in truly meaningful ways I would like Amir to measure the effects. Would you be able to do it yourself, even if you don't own an expensive AP that ASR has?

 

[TimoJ]

do you think chance is reasonably good that distortions/noise would increase enough for it to be audibly discernible, to people with no, to not much hearing loss?

Probably not. Then again, people here tend to obsess over SINAD. Is it actually audible beyond a certain, fairly low value? Enabling Dirac probably reduces SINAD by about 10 dB

 

[Mort]

Enabling Dirac probably reduces SINAD by about 10 dB

You have SINAD chasing shaming syndrome

 

[peng]

Probably not. Then again, people here tend to obsess over SINAD. Is it actually audible beyond a certain, fairly low value? Enabling Dirac probably reduces SINAD by about 10 dB

We shouldn't have to guess.