Audiocontrol Hyperion Processor

[PGAMiami]

Regarding the Trinnov Altitude CI, it can process all incoming audio at its native sample rate up to 24/192. However the Dante output is downsampled to 48kHz because it needs to be for compatibility with all incoming sample rates as Dante must be fixed to a single rate. However if you use the AES/EBU outputs, it’s possible to get the full bitperfect SRC free in-out signal path up to 24/192. Don’t expect this from any other processor that isn’t Intel processor powered. As it take a significant amount of processing power to achieve this.

Taking this back to the Audio Control products, they claim that the ARM processors they are using have x86, aka Intel, processing power. Let’s hope they put the processing power to good use. If I understand the math correctly, to achieve the same level of resolution down to 1hz or 2 which you need to optimally DSP bass, you need to double the number of taps whenever you double the sampling rate. SRC done correctly with 32 bit floating point math should be inaudible to humans. So the decision to keep everything at 48khz seems to make sense in terms of optimizing performance if processing power is a constraint. That said, it would be “nice” and elegant even if not audible to have Dirac operating at the native sample rate of the source if processing power is abundant. I expect the 32 bit floating point math that preserves the 24 bit data is the more important factor, and I believe even basic AVRs can do this.

 

[Kal Rubinson]

Regarding the Trinnov Altitude CI, it can process all incoming audio at its native sample rate up to 24/192. However the Dante output is downsampled to 48kHz because it needs to be for compatibility with all incoming sample rates as Dante must be fixed to a single rate.

Ravenna is not so constrained.

 

[Beast76]

Last I had HTPC was in the elder ages. Actually remember it quite vividly as it broke some speed records while on the balcony of my NY flat back in 2001. Was -25C so it was a really good run.

Enjoying my ART on simple AV 10 - really no reason for PEQ or after the fact tone adjustments in my use case. Just perfect the way it is.

While I’m happy you’re happy - I also know you’d be happier with tone controls after Dirac. All content is different and when you can make very quick adjustments to increase/decrease bass and treble frequencies based upon the content and the volume you’re listening at - it’s a game changer. Necessary? No. Very nice - absolutely.

 

[Highfidelity]

Ravenna is not so constrained.

Trinnov’s Ravenna implementation is software based and also downsampled to 48kHz. And not near as good as their Dante. It has more latency and no leader or Grandmaster clocking capability. Where their Dante implementation is the super solid hardware based card here: https://www.digigram.com/products/sound-cards/alp-dante-pcie-sound-card-smart-and-ultra-versatile/

Edit: When using Dante the Trinnov will resample to whatever you have the Dante network set at. So if you have it set to 24/96 it will upsample all audio that’s lower, and play all incoming 24/96 without SRC. However the software for Ravenna and native AES67 for Dolby processors is limited to 48kHz. So they’re forced to resample everything to 48kHz.

Ravenna only follows the input sample rate and switches on the fly when the Merging Zman is used. This isn’t a special feature of the Ravenna protocol itself. It’s a feature Merging implemented in the Zman board only.

 

[Highfidelity]

Taking this back to the Audio Control products, they claim that the ARM processors they are using have x86, aka Intel, processing power. Let’s hope they put the processing power to good use. If I understand the math correctly, to achieve the same level of resolution down to 1hz or 2 which you need to optimally DSP bass, you need to double the number of taps whenever you double the sampling rate. SRC done correctly with 32 bit floating point math should be inaudible to humans. So the decision to keep everything at 48khz seems to make sense in terms of optimizing performance if processing power is a constraint. That said, it would be “nice” and elegant even if not audible to have Dirac operating at the native sample rate of the source if processing power is abundant. I expect the 32 bit floating point math that preserves the 24 bit data is the more important factor, and I believe even basic AVRs can do this.

But that doesn’t mean they’re using processors as powerful as what Trinnov is using.

FIR is never used below 150hz as you end up getting pre-ringing that destroys the sound quality. So taps don’t matter for bass. Since all Atmos content is 48kHz already, the SRC isn’t even used unless listening to stereo PCM or multichannel DTS or DolbyHD recorded at different sample rates.

 

[mglobe]

While I’m happy you’re happy - I also know you’d be happier with tone controls after Dirac. All content is different and when you can make very quick adjustments to increase/decrease bass and treble frequencies based upon the content and the volume you’re listening at - it’s a game changer. Necessary? No. Very nice - absolutely.

Not positive, but I’’m pretty sure from what I read in the user manual, that the AV10 Dirac implementation can store three presets with different target curves, and switch between them. My MiniDSP Flex could do four different presets. And my Trinnov NOVA can do a ton of them. So DRC (or if you prefer DSP) does not limit you to a single target curve. Not exactly the same as tone controls, but also in some ways better.

 

[PGAMiami]

But that doesn’t mean they’re using processors as powerful as what Trinnov is using.

FIR is never used below 150hz as you end up getting pre-ringing that destroys the sound quality. So taps don’t matter for bass. Since all Atmos content is 48kHz already, the SRC isn’t even used unless listening to stereo PCM or multichannel DTS or DolbyHD recorded at different sample rates.

Even my lowly Denon 4800h as an ART pre/pro is sounding great on video. And I haven’t yet installed the Perlisten subs, just 5.0.2 with TADs and KEF speakers. I’m not too concerned on video. If the tonality of a motorcycle, helicopter or bullet is a little off it’s still great. But when I compare the Denon with ART to my RME with a Mitch Barnett 60k tap convolution filter running on Roon, the RME does win, even with closely matched targets curves. I find high def immersive playing on the Denon with Kodi and an Nvidia Shield clunky and very mid Fi compared to the same material in stereo on the RME. But a full Mitch Barnett multichannel system is complicated and I’m not the only one using the system. So this is why I’m looking for a simple solution that will perform like the RME.

 

[Highfidelity]

Even my lowly Denon 4800h as an ART pre/pro is sounding great on video. And I haven’t yet installed the Perlisten subs, just 5.0.2 with TADs and KEF speakers. I’m not too concerned on video. If the tonality of a motorcycle, helicopter or bullet is a little off it’s still great. But when I compare the Denon with ART to my RME with a Mitch Barnett 60k tap convolution filter running on Roon, the RME does win, even with closely matched targets curves. I find high def immersive playing on the Denon with Kodi and an Nvidia Shield clunky and very mid Fi compared to the same material in stereo on the RME. But a full Mitch Barnett multichannel system is complicated and I’m not the only one using the system. So this is why I’m looking for a simple solution that will perform like the RME.

Combine a Trinnov Altitude CI with a Merging HAPI MK3 connected via AES/EBU. And you’ll have native SRC free processing up to 24/192, and arguably the best room correction available. And DAC quality that exceeds the RME.

 

[welwynnick]

Combine a Trinnov Altitude CI with a Merging HAPI MK3 connected via AES/EBU. And you’ll have native SRC free processing up to 24/192, and arguably the best room correction available. And DAC quality that exceeds the RME.

I'm glad you suggested that. That would be my idea of heaven, too. Perhaps next year....

One question though. You mentioned a couple of times that Trinnov offer an SRC-free pipeline, and I guess that's true (for AES/EBU output). We also know that StormAudio use ASRC on both the input and the output, regardless of whether the output is analogue, AES/EBU or Dante. How do you KNOW that Trinnov don't do the same?

 

[PGAMiami]

Putting aside the debate on audibility of SRC or DAC SINAD, key question for me I how does Dirac ART ART MIMO on the Hyperion compare with Trinnov Room Optimizer? Two very different approaches. And I think we would have to separate the discussion on how these perform on bass vs midrange and above.

 

[PGAMiami]

Combine a Trinnov Altitude CI with a Merging HAPI MK3 connected via AES/EBU. And you’ll have native SRC free processing up to 24/192, and arguably the best room correction available. And DAC quality that exceeds the RME.

Yes but not the most friendly system for simple IR remote. No CEC on the Trinnov or even IR volume control on the Hapi. Difference in DACs I expect to be small but differences in ART vs Trinnov DSP would be quite significant

 

[welwynnick]

...... key question for me I how does Dirac ART MIMO on the Hyperion compare with Trinnov Room Optimizer?

Phew. That's been debated endlessly elsewhere, and without resolution in sight.
I think Storm have an operational convenience advantage (Trinnov isn't very useable, slow to start-up, slow to switch, etc).
I think Trinnov have a DSP advantage, using a PC architecture affords more than adequate DSP power.
Storm ART competes with Trinnov wave-forming, but ART seems to be more sophisticated - a 3D solution as opposed a 2D solution (and you don't have to buy a Storm).
Having seen and heard the evidence, I think both ART and wave-forming are rather unnatural and artificial sounding, and perhaps not the real deal-breaker.

 

[Highfidelity]

I'm glad you suggested that. That would be my idea of heaven, too. Perhaps next year....

One question though. You mentioned a couple of times that Trinnov offer an SRC-free pipeline (except where Dante is used, where it's essential). We also know that StormAudio use ASRC on both the input and the output, regardless of whether the output is analogue, AES/EBU or Dante. How do you KNOW that Trinnov don't do the same?

Because Trinnov explains on their website that when AES/EBU is used, no ASRC is used. You get native processing at all sample rates from input to output.

 

[welwynnick]

Because Trinnov explains on their website that when AES/EBU is used, no ASRC is used. You get native processing at all sample rates from input to output.

Thanks.
Got a link?

 

[Highfidelity]

Phew. That's been debated endlessly elsewhere, and without resolution in sight.
I think Storm have an operational convenience advantage (Trinnov isn't very useable, slow to start-up, slow to switch, etc).
I think Trinnov have a DSP advantage, using a PC architecture affords more than adequate DSP power.
Storm ART competes with Trinnov wave-forming, but ART seems to be more sophisticated - a 3D solution as opposed a 2D solution (and you don't have to buy a Storm).
Having seen and heard the evidence, I think both ART and wave-forming are rather unnatural and artificial sounding, and perhaps not the real deal-breaker.

In order to compare 2 setups accurately, both setups must have the same dacs, amps, speakers and levels matched within 0.1db. You can’t listen to one room correction on a $1500 receiver, and compare with 64 bit floating point room correction on PC with dacs that have -122dB THD+N connected to Purifi amps and say Dirac sounds artificial in comparison. There’s 1000 other variables in the equation that could be causing the artificial sound.

If you were to say I heard a Storm EVO connected to a Merging HAPI Mk3 DAC via Dante, followed by Purifi 1ET6525SA based amps, connected to a full speaker setup from KEF or whatever brand. Then did critical listening tests. Then right after you swapped in the Altitude CI connected to the same chain via Dante, and ensured 0.1db level matching from the Storm setup. Then did critical listening. And after you looped the Dante from the CI with room correction disabled, though a PC and used Acourate or Audiolense based room correction instead, level matched. Then had a listening session. Only then might I consider listening to your subjective thoughts between the room correction formats.

Regarding the Trinnov CI, the new 3D mic is a single Ethernet connection powered by PoE. And you only need a single sweep to calibrate every channel all at once. It’s by far the fastest way to correct a room.

 

[Highfidelity]

I

Thanks.
Got a link?

I don’t have time to search for it now. But it’s there somewhere.

 

[welwynnick]

If you were to say I heard a Storm EVO connected to a Merging HAPI Mk3 DAC via Dante, followed by Purifi 1ET6525SA based amps, connected to a full speaker setup from KEF or whatever brand. Then did critical listening tests. Then right after you swapped in the Altitude CI connected to the same chain via Dante, and ensured 0.1db level matching from the Storm setup. Then did critical listening.

I'm only aware of one instance of where anything like that has ever happened, and the result was of such great value to the dealer, that he kept it secret.

 

[Kal Rubinson]

Trinnov’s Ravenna implementation is software based and also downsampled to 48kHz.

AFAIK, that applies to all the Trinnovs other than the CI.

Combine a Trinnov Altitude CI with a Merging HAPI MK3 connected via AES/EBU. And you’ll have native SRC free processing up to 24/192, and arguably the best room correction available.

Part of my plan.

 

[Highfidelity]

I'm only aware of one instance of where anything like that has ever happened, and the result was of such great value to the dealer, that he kept it secret.

I do things like that all the time. Soon I’ll be doing it with the Storm EVO, Trinnov CI, and Hyperion DPR-16.

At this point I’m thinking the winning setup will be the Trinnov CI over Dante with Dante network set at 24/96. This way only 44.1 and 48 will be upsampled. And SoX based 64 bit floating point DSP SRC that Trinnov uses is very transparent upsampling to 24/96. And native 24/96 content will be sent bitperfect through the whole chain to DAC.

I also have a Appsys Flexisys AES with Dante daughter card. So I can test the AES outs too and compare. But it’s really not needed because I can compare 48kHz bitperfect to 48kHz upsampled to 96kHz simply by changing the sample rate in Dante controller.

 

[Highfidelity]

AFAIK, that applies to all the Trinnovs other than the CI.

Part of my plan.

From their website: