• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. 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!

Why can't audio industry standardize on a common digital audio "hdmi"?

win

Senior Member
Joined
Jun 17, 2020
Messages
430
Likes
432
Location
Irvine CA
thats not relevant. You can have 32 channels on usb2. You can have hundreds of channels on a 1gb connections. Latency is important for studios, not for consumers. A convolution filter for room correction will add a lot of latency (~150ms).

there is already too many connectors. Please don’t add a new one.

Non sequitur noted
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
Latency is important for studios, not for consumers. A convolution filter for room correction will add a lot of latency (~150ms).
This is not correct. Audio isn't always just music. Audio chains with high-fidelity requirement are also used in conjunction with video (in HT applications), lip sync is a requirement and puts a limit on latency. Even in consumer audio, if you are doing room distribution digitally, and playing via multiple speakers, you want latency to be controlled to minimize the echo effect.

Parametric room correction filters (the most common type) even done in software on a PC are typically less than 10ms latency. You can indeed construct processing that are orders more in latency but that is certainly not the norm and avoided for HT applications.

You can provide certain latency guarantees in IP based networks but the requirements on the equipment to do so on each device is very high in addition to requirements on the intervening hubs/switches with QoS features out of the control of the device manufacturer (which increases the chances of the unit being perceived by the customer as faulty). So, not a good replacement for short-haul interconnects, especially in consumer households (other than households with network engineering skills).

there is already too many connectors. Please don’t add a new one.
This is totally missing the point. The need for a universal connector is precisely because there are too many connectors fragmenting applications that put a burden on device manufacturers and limit emergence of new applications. So the need for the industry is to adopt a universal connector that makes all those different connectors unnecessary AND simplifies transceivers AND encourages new entrants and applications. So, it will only happen if the industry agrees that such a universal connector will solve the problem of too many connectors, not just add one more.

This is all covered in my post on problem statement and value props, so I am not going to repeat them here.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire
This is not correct. Audio isn't always just music. Audio chains with high-fidelity requirement are also used in conjunction with video (in HT applications), lip sync is a requirement and puts a limit on latency. Even in consumer audio, if you are doing room distribution digitally, and playing via multiple speakers, you want latency to be controlled to minimize the echo effect.
You are confusing low latency with synchronisation. You can have either one without the other.
 

maverickronin

Major Contributor
Forum Donor
Joined
Jul 19, 2018
Messages
2,527
Likes
3,308
Location
Midwest, USA
You are confusing low latency with synchronisation. You can have either one without the other.

There's also gaming. Games always need low latency since you can't just delay the video like you can with a movie.
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
You are confusing low latency with synchronisation. You can have either one without the other.

Not necessarily. The only way to synchronize with varying latency is to bring everything down to the highest latency and that is not always acceptable or even necessarily possible.

Synchronization requires latency guarantees. If it is just music, then latency levels don't matter (within reason), music just starts a bit later. If it is video+audio then things get a little bit complicated especially if the video rendering and audio rendering is done by separate entities.

So, for example, take a media player on a PC doing software rendering of both video and audio (where the latency in this case is from the decoding/rendering). Since both are under its control, it can do synchronization between audio and video by appropriate delaying/buffering.

But consider an application in which a TV streamer app renders video on the screen and sends the corresponding audio stream to be played on a soundbar or through an AVR (Optical, ARC, etc). The only way that works is with minimizing latency since they are rendered by different entities.

You can possibly build a synchronization protocol to handle such cases but that only increases the complexity and is certainly not part of any existing standard for inter-connects.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire
Synchronization requires latency guarantees.
Synchronisation requires bounded jitter and a large enough buffer.

You can possibly build a synchronization protocol to handle such cases but that only increases the complexity and is certainly not part of any existing standard for inter-connects.
Oh, but it is. HDMI comes to mind.
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
Synchronisation requires bounded jitter and a large enough buffer.
I have already addressed this above when different entities are rendering video/audio. Both when it is possible and when it isn't. I don't think you are understanding the issue.
Oh, but it is. HDMI comes to mind.
You are still not getting it. The sync of audio to video in HDMI stream can only happen when the same entity is rendering both video and audio. There is no protocol to have two entities co-operate with their own buffers by exchanging sync information. This is what would be required if video and audio were being rendered by different entities (say a TV and an AVR, the most common). Or you can provide a manual sync setting in one or both so users can adjust it manually or the Lip Sync feature in HDMI. This has limitations on latency (or don't allow for variable latency within the same stream). In either case (including eARC), only processing/rendering latencies are accounted for, not network latency (outside the control of the devices invvolved which would require some out-of-band synchronization protocol in real time).

I am just going to drop it here with you because these short assertive statements that don't show evidence of understanding the issues raised is not something one can meaningfully discuss in a two-way. So, let us agree to disagree.
 
Last edited:

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire
The sync of audio to video in HDMI stream can only happen when the same entity is rendering both video and audio. There is no protocol to have two entities co-operate with their own buffers by exchanging sync information. This is what would be required if video and audio were being rendered by different entities (say a TV and an AVR, the most common).
Yes, and HDMI provides the necessary information. Maybe you should read the existing specs before demanding a new one.
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
Yes, and HDMI provides the necessary information. Maybe you should read the existing specs before demanding a new one.
People who do arguments with "if only you read or do your research.." seldom have the understanding. Have already gone through one bout of that earlier with another poster. So, like I said, I beg to disagree from you and take exception to you telling me what I should or should not do. Good bye.
 

StefaanE

Addicted to Fun and Learning
Forum Donor
Joined
May 12, 2020
Messages
528
Likes
929
Location
Harlange, Luxembourg
You can send a file to another device for example using AirPlay and the destination can decode and play it even in real time (because encoded files can be streamed too) if all the processing is within the same box until conversion to analog. So an all-in-one AVR can and do act as a network streamer that can receive MP3 files that it can decode and play. That is not what we are talking about here.
Exactly. The AVR contains a computer that's capable of receiving data through a network, and producing analogue signals for the speakers. It doesn't matter which network technology is used, as long as the speed is enough to supply the AVR's DACs with the required bitstream. But note that AirPlay is proprietary and includes DRM, so it's not a suitable stack for universal interconnection.
When you separate out the digital processing that happens between decoding and conversion to analog in external boxes from multiple vendors - DACs, surround processors, zone distribution, room eq, etc. nothing necessarily to do with analog in this chain, you are doing real time inter-connects of the digital audio stream from the decoder stage. There are no "files" here as you are thinking.
That's the LPCM world you're talking about, and that's why you're thinking of connectors and cables. I was suggesting to think in terms of transferring data between intelligent devices instead of transferring LPCM streams between relatively dumb devices, no more. I think today most music and video is consumed through smartphones or devices using smartphone technology. More and more speakers are smart, and more and more manufacturers offer networked sound (Yamaha MusicCast, D&M Heos, etc.). Why not jump on that bandwagon with an Open Source stack and high-quality smart speakers?
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
Exactly. The AVR contains a computer that's capable of receiving data through a network, and producing analogue signals for the speakers. It doesn't matter which network technology is used, as long as the speed is enough to supply the AVR's DACs with the required bitstream. But note that AirPlay is proprietary and includes DRM, so it's not a suitable stack for universal interconnection.

I see where you are coming from.

The interconnects I am talking about aren't to send from an iPad to an AVR. We are talking about two very different ecosystems.

Take, for example, the connectivity between a TV and an AVR for the TV to send the extracted audio from the video stream it is rendering to a sound bar (to understand this get into the next level of detail of what is being transfered - the TV receives and MP4 stream of video). Now figure out what needs to do to send the sound out (in terms of formats and encodings) to a sound bar from that stream to use a network and how it would enforce sync. Please get to the next level of detail on what actually happens between the two, what is the format (including encoding) of the stream going from the TV to the soundbar in this scenario? You will realize that this doesn't fit your model

Or take a stand-alone streamer that would want to use an external DAC like Okto or Topping because they are better than what is in an AVR to do the digital to analog conversion but we need the AVR to get the networked stream (say a streamed MP3 file) and use its codecs because the Okto and Topping are not designed to have all the codecs ever needed (that would destroy the case for having such devices at any reasonable price because they all would have to license the codecs). This doesn't fit your model either.

You seem to be saying all of those are "dumb" devices and so make all of them be able to take a mp3 file instead over the network, then you would not have those devices at all or everything would turn into an all-in-one receiver (at least for digital stages). This is not a realistic scenario in my opinion and so I beg to leave me out of going down that path.

Why not jump on that bandwagon with an Open Source stack and high-quality smart speakers?

Because what you have in mind is an integrated all-in-one solution of a single device that takes an encoded music file stream and does everything it needs to within itself. It is possible in some simple use cases of using a smartphone to listen to a MP3 file. But when the requirements go into requiring a better DAC than what a smartphone has or you want a surround sound processor that can play through a multi-channel speaker system, etc., then either you have to assume your smartphone acquires all those capabilities or it is talking to a single box that does all of these things and so no digital interconnects are needed (it is done via i2s buses inside the box which is fine because there is already a relatively universal connector there!).

I am not dismissing your model in the simpler use case. It is happening now. But I don't see its relevance for the above use cases where everything isn't happening in one box, as you assume.

You are not talking about a new form of interconnects but a different ecosystem of devices themselves. I suggest you start a thread separately on how that would work to replace all of the current "dumb" devices. The interconnects are not relevant in that world but that new ecosystem design is not relevant to this existing world.

So I am going to stop here with your line of thinking. Thanks for your contribution.
 

StefaanE

Addicted to Fun and Learning
Forum Donor
Joined
May 12, 2020
Messages
528
Likes
929
Location
Harlange, Luxembourg
I see where you are coming from.

The interconnects I am talking about aren't to send from an iPad to an AVR. We are talking about two very different ecosystems.
A last question if I may. When you’re talking about using a better DAC, you’re leaving the digital world, is it not? Most DACs are stereo devices, so a new multi-channel digital connector would also require a 5+1 DAC, I surmise. Do you think there is a market for such devices, given that the ‘phoolish money is spent on 40k power leads by people who believe in the superiority of stereo?
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
A last question if I may. When you’re talking about using a better DAC, you’re leaving the digital world, is it not?
A DAC by definition takes digital input and outputs analog. So, it has digital connections for input and analog connections for output.
Most DACs are stereo devices, so a new multi-channel digital connector would also require a 5+1 DAC, I surmise.
No. Stereo DACs typically take USB inputs amongst other connector tyoes which can be carry any number of channels from 1-16 or more. They process whichever channels they are capable of. Look at reviews of Topping D10 and the Okto 8 Pro two of the top rated DACs here. The latter is a multi-channel DAC used for multiple purposes.
Do you think there is a market for such devices, given that the ‘phoolish money is spent on 40k power leads by people who believe in the superiority of stereo?
Home Theater is a big market for multiple channel DACs which is why all major DAC chip makers have multi-channel DAC chips. Almost all AVRs for the past decade have had multi-channel DACs in them. It has nothing to do with who spends foolishly for what. Different market segments.

There are very few multi-channel external DACs, part of the reason is exactly what creates a motivation for this proposal that multi-channel transmission connectors are owned by the HDMI mafia with expensive and restrictive licenses while the rest of the industry is fragmented over USB, AES/EBU, S/P-DIF, etc.
 

pierre

Addicted to Fun and Learning
Forum Donor
Joined
Jul 1, 2017
Messages
962
Likes
3,048
Location
Switzerland
This is not correct. Audio isn't always just music. Audio chains with high-fidelity requirement are also used in conjunction with video (in HT applications), lip sync is a requirement and puts a limit on latency. Even in consumer audio, if you are doing room distribution digitally, and playing via multiple speakers, you want latency to be controlled to minimize the echo effect.

Parametric room correction filters (the most common type) even done in software on a PC are typically less than 10ms latency. You can indeed construct processing that are orders more in latency but that is certainly not the norm and avoided for HT applications.

Let's do it the other way around. Tell me your max latency, I can tell you if TCP/IP does it or not with cheap switches and/or on wifi. And if you use a rj45 cable point to point, then the point is moot.

You can provide certain latency guarantees in IP based networks but the requirements on the equipment to do so on each device is very high in addition to requirements on the intervening hubs/switches with QoS features out of the control of the device manufacturer (which increases the chances of the unit being perceived by the customer as faulty). So, not a good replacement for short-haul interconnects, especially in consumer households (other than households with network engineering skills).

See above. if we are talking a few ms then any switch at 30$ will make it.

This is totally missing the point. The need for a universal connector is precisely because there are too many connectors fragmenting applications that put a burden on device manufacturers and limit emergence of new applications. So the need for the industry is to adopt a universal connector that makes all those different connectors unnecessary AND simplifies transceivers AND encourages new entrants and applications. So, it will only happen if the industry agrees that such a universal connector will solve the problem of too many connectors, not just add one more.

This is all covered in my post on problem statement and value props, so I am not going to repeat them here.

Industry doesn't want a common port. They do sell bunch of cables and adapters at high margins (Apple is the best example). And most of them don't want to make life too easy to new entrants. On PC they could all have settle on thunderbolt years ago but they didn't because they compete with each others (we can dream that usb4 will fix that with enough bandwidth and enough power).

Other reasons: hdmi is bulky and for example will not work well on a mobile. same with RJ45. USB4 is more likely to succeed technically because it also carries power (PoE). But for the reasons above, it will not. You would need a large set of regulators to force the large company to agree on something like this.
 

andymok

Addicted to Fun and Learning
Joined
Sep 14, 2018
Messages
562
Likes
553
Location
Hong Kong
I sure hope some random person can't listen in on my VoIP phone calls simply by joining a multicast group.

I was referring Video-over-IP, ST 2110, where AoIP AES67 being a sub-set of it :p
 
Last edited:

StefaanE

Addicted to Fun and Learning
Forum Donor
Joined
May 12, 2020
Messages
528
Likes
929
Location
Harlange, Luxembourg
Home Theater is a big market for multiple channel DACs which is why all major DAC chip makers have multi-channel DAC chips. Almost all AVRs for the past decade have had multi-channel DACs in them. It has nothing to do with who spends foolishly for what. Different market segments.

There are very few multi-channel external DACs, part of the reason is exactly what creates a motivation for this proposal that multi-channel transmission connectors are owned by the HDMI mafia with expensive and restrictive licenses while the rest of the industry is fragmented over USB, AES/EBU, S/P-DIF, etc.
Thanks for the references, I looked at the Okto DACs. Super nice kit, but the stereo version has only two output channels. The Pro has eight XLRs, but needs to be connected to a device that can send it 8 digital channels. You’re looking at studio equipment here, not consumer grade AV preprocessors — these things have only analogue outputs. My problem is not the connector, but finding a device that will take multi-channel source material such as an SACD or Blu-ray audio disk and give me 5.1 digital audio outputs. The best I can do is take my Marantz 5013 and connect its analogue pre-outs to my active speakers, which start by re-digitising the signal before using their own DSPs to distribute the signal over their 4 driver amps. With stereo material, I can keep the chain digital until the final amplification in the speakers.
So yes, I’m on your side WRT having access to multi-channel digital audio, but I don’t see how a different connector would achieve that.
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
Let's do it the other way around. Tell me your max latency, I can tell you if TCP/IP does it or not with cheap switches and/or on wifi. And if you use a rj45 cable point to point, then the point is moot.
It is not the question of cost of the switch/hub. It is the cost of the Oktos and the Toppings having to incorporate a full IP stack in their boards, the cost of having a discovery process on the network to figure out who to send it to or having the user configure it with IP addresses, it is having to depend on an average consumer having configured their router and switch QoS to ensure that they meet the latency standards required for their equipment preferably with zero config on part of the user, etc.

To a hammer everything looks like a nail. IP network is not a substitute for an equipment direct inter-connect. In the same way a network is not a substitute for all USB interconnections. It is not a hard concept.

If you want to use a rj45 cable point to point as the solution then you have just proposed a specific universal connector for the industry to adopt (and a heavyweight stack to do what they need to do). :)

Industry doesn't want a common port. They do sell bunch of cables and adapters at high margins (Apple is the best example). And most of them don't want to make life too easy to new entrants. On PC they could all have settle on thunderbolt years ago but they didn't because they compete with each others (we can dream that usb4 will fix that with enough bandwidth and enough power).
Oktos, miniDSPs, Toppings, Schiits. RMEs, Motus, just to name a few aren't in it for selling expensive cables. They do suffer from having to provide multiple types of interfaces and dongles and adapters because of fragmented industry connectors.

Apple does it because it has a monopoly in its ecosystem. But that is not the industry we are talking about.
Other reasons: hdmi is bulky and for example will not work well on a mobile. same with RJ45. USB4 is more likely to succeed technically because it also carries power (PoE). But for the reasons above, it will not. You would need a large set of regulators to force the large company to agree on something like this.
Wrong premise, you don't need large companies to agree to this just enough of the smaller players (within consumer and pro-audio groups) doing specialized boxes to start with.

Skepticism is ok, I understand. Simple naysaying, I don't care much for as valid. Startups depends on there being a lot of naysayers discouraging others allowing them to disrupt big players. :)
 
OP
Vasr

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,923
So yes, I’m on your side WRT having access to multi-channel digital audio, but I don’t see how a different connector would achieve that.

The key phrase is "universal connector adopted by industry" not someone doing a different connector for existing equipment. Understand the difference?

With the former, your SACD player or your PC soundcard or your Blu-ray player will have a universal port (whatever technology/protocol it is based on), the Oktos, Schiits. Motus and Toppings will have the same connector instead of a mess of optical, coax, AES, etc. etc. As a consumer you plug your SACD player to the Okto multi-channel witha single cable and your Okto multi-channel to a multi-channel amp and you are set with multi-channel music. No need for expensive all-in pre/pro boxes just to do that. If you want to introduce room correction for your multi-channel music, a miniDSP Dirac will have the same connectors for input and output and you plug it inline between your SACD player and the Okto DAC and you are done with the connectivity. These are just examples. But that is the convenience and usability that industry adopting a universal connector will bring.

To do the same now, you need to figure out how to get your SACD player to output to a USB so you can connect it to the Okto or convert optical to a AES connector (this is why you think Okto is a pro-audio equipment not because of what it does but what connectors it has, think about that).
 
Top Bottom