Possible solution for DTS/Dolby/Atmos+eARC+HDCP to AES/EBU - via Dante? (for Okto DAC8PRO etc)

[ChrisUpNorth]

Does anyone know whether the SW42DA can directly decode from an Apple TV HDMI output (or like whether there needs to be an AVR in between them)? It would be interesting to know from someone who has first-hand experience, thanks.

 

[ChrisUpNorth]

Does anyone know whether the SW42DA can directly decode from an Apple TV HDMI output (or like whether there needs to be an AVR in between them)? It would be interesting to know from someone who has first-hand experience, thanks.

I've since heard back from Aiden at Blustream who confirms that the Apple TV can plug directly into the SW42DA, so no need for an AVR at all; the HDCP applies to the video stream only so the device can extract the full multichannel audio stream.

 

[welwynnick]

I've since heard back from Aiden at Blustream who confirms that the Apple TV can plug directly into the SW42DA, so no need for an AVR at all; the HDCP applies to the video stream only so the device can extract the full multichannel audio stream.

I'm pretty sure that HDCP applies to audio as well as video, but maybe that's only high resolution lossless audio from Blu-ray?
I think Apple TV and other streaming services use Dolby Digital plus.
If that's the case, I wonder if the Blustream will decode Dolby true HD etc?

 

[ChrisUpNorth]

I'm pretty sure that HDCP applies to audio as well as video, but maybe that's only high resolution lossless audio from Blu-ray?
I think Apple TV and other streaming services use Dolby Digital plus.
If that's the case, I wonder if the Blustream will decode Dolby true HD etc?

According to their literature...

Supports all known HDMI audio formats including Dolby TrueHD, Dolby Atmos, Dolby Digital Plus and DTS-HD Master Audio transmission

I'm pretty much ready to pull the trigger on a SW42DA. I know it is restricted to HDMI v2 (not 2.1) but I'm not interested in the benefits of 2.1 which I believe are mostly beneficial for gaming. I've currently got a 4-speaker multichannel setup working really well with Roon & Dante so my next logical step is to introduce a HMDI feed from Apple TV so I can listen to their Dolby Atmos music and videos (albeit in 4-channel). The Nu-Prime looks interesting but (a) it's not available yet and (b) the room-correction is an uknown, whereas with the Blustream I would use DLRC on one of my macs/roon endpoints.

 

[darth]

For folks in the US, where are you buying the Bluestream from? I only see it listed at HDTV supply, and that too at a large markup over EU prices.

 

[chelgrian]

I'm pretty sure that HDCP applies to audio as well as video, but maybe that's only high resolution lossless audio from Blu-ray?
I think Apple TV and other streaming services use Dolby Digital plus.
If that's the case, I wonder if the Blustream will decode Dolby true HD etc?

HDCP applies to the entire data stream what they really mean is that the licensing restrictions on outputting un-encrypted data apply to video and were relaxed not to apply to channelised decompressed digital audio data, which they originally did. They do still apply to the original compressed audio stream.

 

[kfrancis]

Here's an update to the list with the Apantac, AudioControl, Blustream and Nuprime added:

View attachment 436339

Interestingly, the AudioControl X7S and X9S are now listed as discontinued. Presumably the APR-16 and DPR-16 will replace them.

Small corrections to this chart re the Arvus H2-4D (source: I own one):

• for Dante it supports up to 192 kHz if you really want it to
• but for AES67 it's limited to 48 kHz, which I think is the case for all hardware-based Dante devices – the H2-4D has a Dante Broadway in it
• It can output 9.1.6, but strangely, it cannot output 7.1.6
• It looks like they've raised the price to 5,350 USD as well – it was 4,990 USD when I bought it last November

I'm happy to answer any questions anyone has about the H2-4D. It mostly does what it says it does quite well.

But it definitely has one very annoying bug I feel compelled to caution about: when the HDMI source is outputting PCM stereo, as soon as there is silence in the output, the H2-4D almost immediately switches its detected codec to simply "N.E." (noise equivalent?). The problem is that when there is output from the source again, it takes the typical second or so to detect PCM2/0 and start outputting again. This manifests as, for example, missing the first ~second of audio of any YouTube video, or Apple Music stereo track, or after unpausing any stereo source, etc. This can even happen in the middle of a track or video, if there is "absolute" silence for a second or so. None of the other codecs seem to exhibit this problem. The codec detection and switching delay to, e.g., Atmos or DTS:X, actually seems noticeably faster than, say, the JBL SDP-58 I owned previously. But obviously a lot of content will always remain stereo, so this bug is a source of daily annoyance.

Another word of caution: Arvus, the company itself, is also, let's say, challenging to communicate with. I'm hopeful all these new, much cheaper devices can execute better, technically and otherwise.

 

[ChrisUpNorth]

I got my SW42DA from Futureshop in the UK this weekend, really happy with it so far. Whilst I'm waiting for my Hapi to arrive, I've got the SW42DA Dante outputs going to two separate Macs each running Dante Virtual Soundcards - one for the front speakers, the other for the surrounds. Everything works as expected with the Apple TV watching movies and listening to spatial audio - there have been no dropouts and no latency issues; my Dante network is running at 96 kHz. My next project is to send the SW42DA outputs to one of the Macs for DLRC.

 

[welwynnick]

Small corrections to this chart re the Arvus H2-4D (source: I own one):

• for Dante it supports up to 192 kHz if you really want it to
• but for AES67 it's limited to 48 kHz, which I think is the case for all hardware-based Dante devices – the H2-4D has a Dante Broadway in it
• It can output 9.1.6, but strangely, it cannot output 7.1.6
• It looks like they've raised the price to 5,350 USD as well – it was 4,990 USD when I bought it last November

View attachment 438371View attachment 438370View attachment 438372

I'm happy to answer any questions anyone has about the H2-4D. It mostly does what it says it does quite well.

But it definitely has one very annoying bug I feel compelled to caution about: when the HDMI source is outputting PCM stereo, as soon as there is silence in the output, the H2-4D almost immediately switches its detected codec to simply "N.E." (noise equivalent?). The problem is that when there is output from the source again, it takes the typical second or so to detect PCM2/0 and start outputting again. This manifests as, for example, missing the first ~second of audio of any YouTube video, or Apple Music stereo track, or after unpausing any stereo source, etc. This can even happen in the middle of a track or video, if there is "absolute" silence for a second or so. None of the other codecs seem to exhibit this problem. The codec detection and switching delay to, e.g., Atmos or DTS:X, actually seems noticeably faster than, say, the JBL SDP-58 I owned previously. But obviously a lot of content will always remain stereo, so this bug is a source of daily annoyance.

Another word of caution: Arvus, the company itself, is also, let's say, challenging to communicate with. I'm hopeful all these new, much cheaper devices can execute better, technically and otherwise.

Hi kfrancis, thanks very much for your input.
I took my information from the Arvus website, but I may have used H1D specs for the H2-4D as well.
Incidentally I noticed that the H1D isn't accessible from their home page any more? An oversight perhaps?
Can you tell us what's the rest of your system?
Since you were using an SDP-55 before, I presume you're using Dante?
Did you get any benefits from using the Arvus instead of the JBL?
Could you elaborate on "it supports up to 192 kHz if you really want it to"?
When it comes to comms and support from Arvus, I think yours isn't exactly a lone voice....
Regards

 

[chelgrian]

Hi kfrancis, thanks very much for your input.
I took my information from the Arvus website, but I may have used H1D specs for the H2-4D as well.
Incidentally I noticed that the H1D isn't accessible from their home page any more? An oversight perhaps?
Can you tell us what's the rest of your system?
Since you were using an SDP-55 before, I presume you're using Dante?
Did you get any benefits from using the Arvus instead of the JBL?
When it comes to comms and support from Arvus, I think yours isn't exactly a lone voice....
Regards

The H1D appears to be vapourware it’s been ‘coming soon’ for over a year. The pages are still there but not linked. I think there were some beta units shipped at some point in 2024 so presumably the feedback from that meant they have not yet gone in to production.

The Blustream unit can be bought and the Nuprime unit is in pre-order so we’ll see if they can deliver.

 

[kfrancis]

Can you tell us what's the rest of your system?

Oh boy. It's arguably nuts, but here goes:

• 7.2.4 layout with Genelec 8351b for LCR, 8341a for side and rear surrounds, 8331a for heights, 7360a subs
  • the rest of this is fallout from a (perhaps misplaced) desire to keep the signal path digital thru to the speakers
• Apple TV and (recently) Kaleidescape Strato V into AVPro Edge 4x2 HDMI Matrix outputting to TV and to
• Arvus H2-4D to handle audio decoding/rendering, outputting over Dante (96 kHz, 24 bits) to
• DAD Core 256 to do some mixing/summing of channels
  • mainly, subscribe to the Arvus outputs and output the sum as a bass management channel [1]
• Netgear M4250 PoE switch
  • this line of switches has built-in QoS profiles specifically designed for e.g. Dante
• Dante AVIO AES3 adapters plugged right into about half of the speakers [2]
  • the others are connected to an AES pass-thru of one of those
• Mac Mini M4, primarily [1] to handle volume control
  • Genelec GLM software is running here
  • GLM exposes its volume control (and more) over MIDI
  • Software I wrote (in python) pretends to be a Sony receiver listening on a TCP port for volume, mute, power commands from Control4 and then send the corresponding MIDI commands to GLM
  • Previously, this software used a Pulse-Eight HDMI-CEC USB adapter to do the same but with CEC instead of Control4, which was ok but required some gross modifications of libcec and was just not as reliable – my family is not interested in why the volume control is broken!

[1] The Core 256 is also connected to the Mac Mini as a Thunderbolt audio interface. The outputs of the Arvus and the Mac Mini's own 7.1.4 output are summed/switched and the result is output again as discrete Dante channels, and those channels are what the speaker AES adapters subscribe to. That sum is also summed via a 7.1.4 fold down (excluding the LFE channel) to mono and output as a bass management channel, which the subs subscribe to on their AES adapter's channel A (with the LFE channel subscribed on channel B).

[2] If I had it to do over again or wanted to "simplify", I might get a DAD AX Center with an AES expansion card, or just a Genelec 9401a, which didn't exist at the time, use all AES3 cabling and skip the switch and Dante adapters. Most Dante/AES67 devices have two network ports for either redundancy or that act as a tiny switch, so this would work.

Tbh, I maybe should've just bought a StormAudio ISP Evo with AES outputs, but this was/is more "fun", much cheaper – though by no means cheap – and it offers more flexibility: I plan to add some in-ceiling Dante speakers in a few places throughout the house, and now I can simply send them a stereo fold down from the Core 256.

If you don't enjoy this kind of tinkering, I would definitely just get a good processor with balanced outs and go analog to the speakers.

 

[kfrancis]

Since you were using an SDP-55 before, I presume you're using Dante?

I was using Dante with the JBL, yes. As you can probably read elsewhere, it is not a great Dante implementation, unfortunately. It was extremely slow when switching codecs. I think the DSP solution they used is slower than average at detecting it, and then there is some kind of weird clocking reset going on with Dante. Ironically, though, it did not have this issue when just sticking with stereo, unlike the weird issue I described above with the Arvus.

I really wanted the SDP-58 to work out, because it has built-in HDMI switching and HDMI-CEC and other consumer AV processor niceties that pro gear doesn't have. But it was nothing but trouble, really. Firmware updates are weirdly complicated – the option to do it in the management web app just doesn't work – and I'm not sure why my unit wasn't fully updated from the factory in the first place, since I got it years after the most recent firmware update. The DSP hardware itself actually failed, first slowly and then completely over time. JBL fixed it under warranty, but the bad Dante implementation and absurdly slow codec switching that I was hoping would be fixed unfortunately remained. Incidentally, it was very nice to be able to email or call JBL Synthesis and actually get a response. And the warranty process was painless.

Did you get any benefits from using the Arvus instead of the JBL?

The Arvus would have been a little cheaper, if I'd done that from the beginning. It also seems to have a much better Dante implementation. It has a Dante Broadway board in it that supports higher sample rates – the JBL is limited to 48 kHz, which is fine, really, given the content that's available. The Arvus also has a word clock input that's tied into the Dante board, so I have that connected directly to the clock out of the internal clock on the DAD Core 256. This is, of course, totally unnecessary, as Dante's whole thing is precise clocking over the network via PTP, but it's a neat "pro" feature.

I also decided to rely entirely on GLM for volume control, so the digital volume control the JBL was doing was an unnecessary feature at that point. And finally, the Arvus is only 1U instead of ~4U. I have no idea why the SDP-58 with no amps is so huge, but given the DSP chip failure, it maybe has something to do with heat management. The HDMI boards seemed to get very hot.

The Arvus still has annoying delays when switching codecs – I think sadly everything does? – But it's way better than the JBL and probably on-par with say Denon/Marantz. Except, of course, the horribly annoying 2-channel PCM bug described above.

Could you elaborate on "it supports up to 192 kHz if you really want it to"?

I just mean you can set the Dante sample rate to 192 kHz, but I don't think you can actually get any input at 192 kHz over HDMI (2.0), and certainly not from the sources I'm using. I guess this would still be helpful if you wanted to integrate the Arvus into a bigger network of Dante devices that are clocked at 192k for other reasons.

To be honest, I'm not sure where in the Arvus sample rate conversion takes places, but I'm thinking it's in the Dante board. You can't configure sample rate anywhere other than Dante Controller, even if you aren't using the Dante outputs. If you want, say 96 kHz on the AES outputs, you have to tell the Arvus to sync the clock to Dante and configure its Dante device to 96k.

The only other option on the Arvus is to sync to the HDMI clock, which means the output sample rate matches whatever the HDMI input is. In theory this seems great, and I tried it, using the AES3 outputs on the Arvus instead of Dante, but the Genelec speakers took a long while, several seconds, to deal with the AES clock rate change, flashing their LEDs red multiple times in the process, and still seemingly trying to output at the existing rate. The results were not good. I don't know who's at fault there or if that's just par for the course with on-the-fly AES3 clock changes. Regardless, the Genelec speaker internal DSP does SRC on the input to 96 kHz anyway, as far as I know, so it makes sense to me to just send the speakers 96k.

 

[ChrisUpNorth]

If you don't enjoy this kind of tinkering, I would definitely just get a good processor with balanced outs and go analog to the speakers.

I definitely appreciate this sentiment! You start to realise when doing this yourself that the AVR has some very user-friendly built-in features that you simply just take for granted. I just got the massive thumbs-up from my family though that the 4-channel system I just deployed sounds miles better than the Sonos Arc/Sub/Ones it has replaced.

 

[welwynnick]

Oh boy. It's arguably nuts, but here goes:

Well I did ask.
Thank you for the comprehensive replies.
That's one of the most remarkable systems I've ever heard of.

It reminds me of the award winning Grimani / StormAudio demonstration at CEDIA 2023 where they showed an all-digital 11.6.6 system based on the ISP Evo. It was all digital in the sense that it was digital up to the speaker cross-overs, and there was one DAC and power amp for each speaker drive unit. They were boasting how networked audio could simplify complex installations, though ironically the ethernet cables were very short and the scores of speaker cables were very long. I think yours is similar in concept, except you use actual digital active speakers.

I trust you're happy with the sound.

This is an endeavour that I've been pursuing for a long time here and here and here. I started out frustrated with the unreasonable expense of Trinnov, Storm and Lyngdorf processors, and the technical mediocrity of their electrical performance, as measured in these pages. I was determined to find a better system solution with better performance and lower cost. I summarised what I found in the attached Excel workbook. Active speakers with digital inputs and DSP crossovers like you use is an equivalent architecture.

In the absence of good PC solution for Atmos decoding, I focussed on audio processors with multi-channel digital audio outputs, so I could use better multi-channel DACs. My best options were the SDP-55 or the Arvus H1D (though now we also have the forthcoming Nuprime H16-A), and in anticipation of all that I got things in the wrong order, and got an AVID HD IO with 16ch AES inputs and 16ch analogue outputs, plus a AVN-AESIO8R to convert Dante to 16ch AES/EBU.

Along the way, many things dawned on me, including what Trinnov was doing so well. Many people thought it was just room equalisation but done better with more DSP power. However it's not just amplitude equalisation, it's phase equalisation as well, and it can correct what the speakers are doing wrong, as well as the room. Three way speakers with complex crossovers are the worst offenders, including actives. Genelec realised this when they collaborated in a research project with Aalto University, Department of Signal Processing and Acoustics that showed that small time delay / group delay / phase variation was audible. Genelec subsequently released GLM version 4.1 with a new feature called extended phase linearity. A few digital active speakers such as Neumann, Kef and Buchardt are starting to appear with linear phase response, and much better transient response as a result. Trinnov do this at system level.

I noticed that you use 8351b, 8341, 8331a, 7360a, and GLM, which are compatible (the original 8351A is not).
So now I'd like to ask - what version of GLM do you use - and have you tried using Extended Phase Linearity?

 

[uriy-ch]

I see people who use Dante are chatting here.

I want to know your opinion about the demand for power DACs with Dante input.
I recently knew someone who ordered a power DAC assembly that has an optical spdif input and output and a Dante input and output.
It turned out like this, it's a power DAC, three channels of 100 W each at a 4 Ohm load with 32 V power supply.






On the power DAC board there is an ADAU1452 that receives and sends back two stereo I2S channels to Dante, the DSP also receives and sends back spdif.

I don't understand how much such a solution can be in demand by people who use Dante?
I'm interested in your opinion on this issue.

 

[kfrancis]

I trust you're happy with the sound.

I wouldn't dare compare it to one of the million dollar CEDIA theaters, but to me it's been sublime, yes.

I'm extremely fortunate to have a wife who's willing to let the room look like a recording studio!

They were boasting how networked audio could simplify complex installations, though ironically the ethernet cables were very short and the scores of speaker cables were very long. I think yours is similar in concept, except you use actual digital active speakers.

It certainly simplifies things to have all three amps within the speaker enclosure. I can see how that would be pretty challenging thermally with big in-wall speakers, though. I can't recall who, but I have seen some project somewhere whose Dante/AES67 amps were colocated very near the drivers in an open hallway type of space behind the wall, rather than the rack.

The Dante adapters, despite being a little awkward, do simplify the cabling a bit, because running Cat 5e/6 is about as cheap and easy as it gets. But you do still have to supply each speaker power, and the daisy-chained RS485 GLM network is pretty annoying with 13 speakers. I hope Genelec eventually releases studio monitors with built-in Dante or AES67 that also use IP for GLM, like Neumann has done.

So now I'd like to ask - what version of GLM do you use - and have you tried using Extended Phase Linearity?

I have – it's on by default with supporting models. I think it just adds an (imperceptible) amount of extra DSP latency. I'm on GLM 5.1.

On a related topic, GLM phase aligns each subwoofer to only one particular speaker of your choosing, e.g., the closest one. I think Dirac, Trinnov, etc. are doing something far more sophisticated with multiple subwoofers.

 

[Esprit]

I hope Genelec eventually releases studio monitors with built-in Dante or AES67 that also use IP for GLM, like Neumann has done.

Smart IP - Genelec.com

Smart IP is every installer’s dream: a range of networked active loudspeaker systems that deliver exceptional audio, power and sophisticated loudspeaker management via a single CAT cable.

www.genelec.com

 

[kandamrgam]

Smart IP - Genelec.com

Smart IP is every installer’s dream: a range of networked active loudspeaker systems that deliver exceptional audio, power and sophisticated loudspeaker management via a single CAT cable.

www.genelec.com

But those are not a good fit for Atmos setups because one, they don't support GLM, two, they are low powered (meant for museums and cafes)

 

[welwynnick]

They're Power-Over-Ethernet only, so I think they probably show the direction Genelec are going in, rather being any sort of end game.

 

[edechamps]

On a related topic, GLM phase aligns each subwoofer to only one particular speaker of your choosing, e.g., the closest one. I think Dirac, Trinnov, etc. are doing something far more sophisticated with multiple subwoofers.

Does GLM support what is sometimes called "Sound Field Management" (SFM) with multiple subwoofers, i.e. EQ'ing each subwoofer separately in order to achieve flat response at multiple listening positions at the same time? Basically the equivalent of MSO. If not, can individual subwoofers be EQ'd manually so that one can use MSO and then feed the results to GLM?