Missing Metrics - Latency!

[TK_IIVI]

Please clarify what is meant by "converter" here.

ADC / DACs in general?

Interfaces that connect to PCs or other host computing devices?

Or also including standalone DDCs ?

I think the question is probably best answered in reverse; look at what is recommended for use by professional musicians and audio engineers. It's going to be devices by RME, Lynx etc. Could be standalone converters to/from analog with a digital connection to a separate interface (to the computer), or it could be an internal card, or an interface with converters.

Does that answer the question?

 

[john61ct]

Does that answer the question?

No I was specifically asking about consumer devices, as you specified in the text I quoted.

As you stated, the topic here is NOT audio production, but latency can be an issue in several home reproduction-only contexts.

So, I'm trying to understand your use of that term in a consumer audio context.

It can also be applied to passive wiring-only adapters, or power supplies, or...

where latency is a non-issue

 

[drumphil]

I understand, but we already have the Gearspace thread with latency information in it for pro devices. So, is that not sufficient if those are the types of devices we'd use anyway for actual audio production.

Frequently consumer devices (like the SMSL C200 DAC I have) have superior fidelity to "professional" options, and I don't know of many "professional" headphone amps that have that level of fidelity, and/or SPDIF input.

I have spare SPDIF outputs on more than one "professional" audio interface, that could benefit from having such devices connected, and it frees up existing analog I/O for other purposes.

Many "professional" headphone outputs still have 30ohms or greater, headphone output impedance. This was true even for RME up until relatively recently.

The SMSL DAC also has superior analog line output fidelity compared to many "professional" audio interfaces and DAC's. It frequently costs a LOT more to get that kind of fidelity in professional devices.

Professional equipment is defined by what it gets used for as much as the intended market segment, and having this information expands the range of equipment that can be used in professional settings.

 

[Ropeburn]

Frequently consumer devices (like the SMSL C200 DAC I have) have superior fidelity to professional options, and I don't know of any professional headphone amps that have that level of fidelity, and/or SPDIF input. I have spare SPDIF outputs on more than one "professional" audio interface, that could benefit from having such devices connectred, and it frees up existing analog I/O for other purposes.

While that is generally true, it's also a good indicator of what really matters. If you can produce great quality music with "mediocre" performing converters (AD too), it means this level of fidelity is factually needless. You won't benefit because it's already way better than enough.

 

[drumphil]

While that is generally true, it's also a good indicator of what really matters. If you can produce great quality music with "mediocre" performing converters (AD too), it means this level of fidelity is factually needless. You won't benefit because it's already way better than enough.

The AD/DA side is usually less important, but the headphone amp side frequently is, with "professional" headphone outputs frequently still having high (30 ohms or higher, amazingly sometimes much higher) output impedance that would be considred completely unacceptable here, and absolutely has an audible effect.

Most affordable headpones these days are low impedance devices. The AudioTechnica ATH-M50x headphones I got recently, $260 Australian, 38 ohms, have issues with many "professional" devices.

Many professional audio interfaces have SPDIF as an I/O expansion option, but there are far fewer stereo SPDIF DAC's available from those manufacturers, and the ones that do exist tend to be far more expensive relative to their fidelity, compared to something like the SMSL DAC I mentioned.

I don't know of a "professional" standalone combination DAC/Headphone amp.

 

[TK_IIVI]

No I was specifically asking about consumer devices, as you specified in the text I quoted.

Ok, here:

"Frequently consumer devices (like the SMSL C200 DAC I have)"

The post immediately following the one I quoted.

I'm getting the feeling that you're trying to make a point rather than understand what I mean, so if you're trying to make a point please just tell me what it is.

 

[phofman]

IMO there is no reason to be limited to professional devices if one requires low latency. IME it's very often only a matter of software - the drivers and the OS audio stack. Any (even professional) device will have long latency when going through the OS audio mixer. Many consumer devices will have very short reliable latency when running through a driver stack with minimum buffers - e.g. ASIO drivers if available, WASAPI exclusive (avoiding the windows mixer), direct alsa hw:X access on linux. A device would have to be quite unusual to have a large latency by itself (e.g. a linux SBC running USB gadget as input (i.e. a UAC2 async device by itself) + DSP SW with output to some real audio device - large overall latency). But for standard MCU-based USB devices (XMOS, STM32, Atmel) - no reason to expect large latency.

 

[nagster]

This is a remeasurement of the impulse response of the USB input.
The difference from last time is that all browsers on the measurement computer were closed and the LAN cable was unplugged. It does not have wireless LAN functionality. This measurement computer is a general-purpose Windows machine used for everyday purposes.
The IR trend has changed. The reduction in delay may be due to a reduction in the computer's load, or it may not be.
The APx hardware is connected to the measurement computer via USB, but it is not an audio device. The Class Code is "Vendor Specific".
With APx, all analysis is performed on the software side. Simply put, the hardware only performs AD / DA / DD.
The audio signal path is "APx500 software > ASIO > USB > ADI-2 DAC > analog > APx555".
There is some delay and variation somewhere in the process, but in any case, the measurement results are as shown in the screenshot.

 

[john61ct]

I'm getting the feeling that you're trying to make a point rather than understand what I mean, so if you're trying to make a point please just tell me what it is.

No, I was just clarifying, trying to understand what you were saying.

But is that what we're talking about? "doing audio production"? Because if it is then using converters made for that purpose is still the way to go, and as I showed earlier there is a latency test floating around of a bunch of the most common devices out there.

I mean, at some point there has to be utility in the testing that's done otherwise it's just a waste of time, and using "consumer" converters for audio production tasks just isn't recommended by virtually anybody who does actual audio production (at least those that do it for a living)

So all the device types I listed would come under that umbrella term.

Since the "audiophile consumer" devices are better SQ I find it odd for pro's to shun them, where interconnects incompatibility is not a problem.

But I agree with your larger point, it seems all the trouble it would take for @amirm to standardize latency specs isn't worth the trouble

 

[amirm]

Sorry guys, didn't see the notifications that you had tagged me (I get way too many of them mixed with likes).

I glanced through the thread. AP software+hardware is not a real-time system. Far from it. It can sometimes fail to even capture its own data reliably! I often have to screw around with ASIO buffer sizes and such to get it to capture samples reliably. As such, in its normal operation, it simply cannot accurately measure latency. I know, I tried long time ago.

There is however, another mode of operation: acoustic sweep. Here, signal processing is used to match up the input and output. You may have seen my remarks in active speakers on how long their latency is. But even here, I have not confirmed that the reported latency is correct. And at any rate, as noted, it is an end-to-end latency, not just the ADC itself.

If there is interest, I can add acoustic sweep testing to ADC. Alas, I don't get enough of these to matter. Or are you all asking for DACs as well? There, testing is already long so I hesitate to add more to it.

 

[TK_IIVI]

No, I was just clarifying, trying to understand what you were saying.

There is no thumbs-up emoji?... (thumbs up)

Since the "audiophile consumer" devices are better SQ I find it odd for pro's to shun them, where interconnects incompatibility is not a problem.

Well I think it mostly ends up being a matter of balance between feature-sets and quality. A lot of us don't buy with a "single use" in mind even if that ends up being the end result. So for example (and sorry for the digression, but since we're chatting...);

I used to work in studios all the time and also managed the hardware and software in one (this is all music-for-post as well as just post). When that studio was set up we picked the industry standard hardware that went with the industry standard software, i.e. Avid's Pro Tools and Avid's hardware. Nothing else was even on the table for the core of the systems.

The thing about places like that was that if we had a problem we needed support right away. Clients were literally sitting behind you and they paid a lot of money to be there. To my knowledge there was never an alternative from consumer companies where they'd offer required support. Another thing was that consumer device drivers don't have good reputation. And a thing on top of that is that some of these studios might end up running literally 24/7 before the gear is shut down long enough to "cool", for weeks on end. In addition there are several pro-audio communities to draw from, and the vast majority I've interacted with are using non-consumer devices.

When I started working 90% at home during the pandemic, which is now 100%, I upgraded my computer and got a new interface because my old Lynx PCI card was, well, PCI. And when it was time to buy a new one I looked at a) well known brands with very good reputation, b) connectivity that I might need, c) price. MOTU 16A worked out well because it has all the i/o I might need along with expandability and Word Clock to sync if necessary. Now, I ended up not needing a fair amount of what it offers, but it's more about viewing my work environment more holistically, as more than just the box, a potentially larger environment where more gear might get hooked up. There's a level of "safety" in choosing pro gear for the aforementioned reasons.

Anyway... "shun" may be a bit glass-half-empty, it's just a pragmatic view on it based on what we know is most likely to work well.

 

[drumphil]

Sorry guys, didn't see the notifications that you had tagged me (I get way too many of them mixed with likes).

I glanced through the thread. AP software+hardware is not a real-time system. Far from it. It can sometimes fail to even capture its own data reliably! I often have to screw around with ASIO buffer sizes and such to get it to capture samples reliably. As such, in its normal operation, it simply cannot accurately measure latency. I know, I tried long time ago.

There is however, another mode of operation: acoustic sweep. Here, signal processing is used to match up the input and output. You may have seen my remarks in active speakers on how long their latency is. But even here, I have not confirmed that the reported latency is correct. And at any rate, as noted, it is an end-to-end latency, not just the ADC itself.

If there is interest, I can add acoustic sweep testing to ADC. Alas, I don't get enough of these to matter. Or are you all asking for DACs as well? There, testing is already long so I hesitate to add more to it.

Hi Amir. What I am asking for specifically is:

For stand alone DAC/ADC units with SPDIF coax or optical I/O, digital to analog and/or analog to digital latency, which I believe is easy to measure with the AP. Correct me if I'm wrong, but I believed that was really as simple as noting a figure, when dealing with test signals sent from the outputs of the AP, returning directly to the AP, without having to get in and out of a computer somewhere in the process.

For audio interfaces and sound cards, ASIO round trip latency. That can be measured with a simple analog cable from the output back to an input, using something like the Oblique RTL utility.

Oblique Audio - RTL Utility

oblique-audio.com

Connect cable, run utility, note round trip times at smaller buffer sizes, say, 16 (for the interfaces that actually offer it), 32, 64, 128, 256, at 44.1kHz, 48kHz, 96kHz. Any higher than that isn't really relevant to real time performance. Only takes a couple of minutes, even testing that many buffer sizes at multiple sample rates.

That's my idea anyway! You'd have to do it, and others may have other ideas of exactly what should be measured.

There seems to be a bit of a divide, between audio production sites that note claimed fidelity figures, but frequently don't actually test them, but do fairly reliably provide round trip latency test results, and this site where you absolutely get the fidelity information, but no latency figures are provided.

Trying to get all the information in one spot can be quite the exercise when attempting to compared different devices. Frequently you end up with one lot of metrics for one device, and a different lot for another, never having a complete comparable set. It is my hope that this falls under the "science" umbrella of this site.

 

[Mnyb]

The AD/DA side is usually less important, but the headphone amp side frequently is, with "professional" headphone outputs frequently still having high (30 ohms or higher, amazingly sometimes much higher) output impedance that would be considred completely unacceptable here, and absolutely has an audible effect.

Most affordable headpones these days are low impedance devices. The AudioTechnica ATH-M50x headphones I got recently, $260 Australian, 38 ohms, have issues with many "professional" devices.

Many professional audio interfaces have SPDIF as an I/O expansion option, but there are far fewer stereo SPDIF DAC's available from those manufacturers, and the ones that do exist tend to be far more expensive relative to their fidelity, compared to something like the SMSL DAC I mentioned.

I don't know of a "professional" standalone combination DAC/Headphone amp.

Is not RME ADI-2 exactly what you need ?

 

[amirm]

For stand alone DAC/ADC units with SPDIF coax or optical I/O, digital to analog and/or analog to digital latency, which I believe is easy to measure with the AP. Correct me if I'm wrong,

Is this the real application for which you want the latency???

Regardless, AP has to capture and digitize the analog output and that latency may be variable.

 

[drumphil]

Is this the real application for which you want the latency???

That is the figure that tells me whther a standalone ADC or DAC will be suitable for low latency operation when integrated into a recording system. In my case, as a high quality DAC and headphone amp, connected to the SPDIF output on my RME HDSP 9652, or the M-Audio Audiophile 2464 card in the system I use with electronic drum kits playing through virtual drum software.

In the case of DAC's that have some kind of DSP capability, who knows how much latency may be added, on top of what can be determined by just looking at a converter spec sheet. Probably not a problem with that specific DAC, but who knows for others.

Regardless, AP has to capture and digitize the analog output and that latency may be variable.

Post #86 and $89 detail using the digital output module of an AP to measure the spdif to analog conversion latency of an RME ADI-2, and unless I'm misunderstanding something, there does not appear to be any variability in the results using this test methodology.

Is not RME ADI-2 exactly what you need ?

Have a look at the price. $2,000 Australian. If the SMSL C200 will do the job, and not have an adverse effect on the latency performance of the system, that's valuable information. Approximately $1,700 of value in the money saved if the SMSL C200 can achieve similar results. I know the DAC fidelity and headphone amp are good enough that I wouldn't need to spend the extra to get the ADI-2. What I don't know is the latency performance.

I'm not providing these examples just so I get an answer to a specific question, but because I think the answers are more generally useful for people involved in audio production.

 

[MC_RME]

Next, I performed measurements using a different AP (APx525) and a different computer, but the results were basically the same. However, the distribution was between approximately 10ms and 35ms.

On the other hand, changing the volume or muting in the Windows mixer did not change the output of the ADI-2 DAC. The limiter did not activate even with 0dBFS output.

As far as I could tell, when I launched the APx500 software and selected ASIO output, the only way to change the ADI-2 DAC output on the computer was through the APx500 software.
Based on these points, it seems that the Windows mixer is not being used. However, the fluctuations are reproducible.

I don't know anything about the internal workings of the APx500 software. Since it's an APx500 and RME combination, MC_RME might know the truth.


@MC_RME,
Could you please tell us about these fluctuations?

Using USB/ASIO on one side and a fixed hardware analysis on the other will give you these variations. ASIO or not - you are starting a software playback, and the OS will make sure that there is no fixed starting point in time for it (Windows/Mac are no real-time OS). ASIO used as both I/O can take care of that as record and playback are always started simultaneously, but here you remove the record part.

Whenever it is about timing you should use the digital output of the AP - that one is fully synced to the AP's digital and analog input, and thus allows to measure hardware latency in an I/O loop - as confirmed in your measurements.

 

[nagster]

I understand. Thank you very much for taking the time to reply.

 

[Dan Lavry]

While I think audio fidelity is well covered in the measurements reported in reviews here, as someone who works in the field of audio production, there is one thing missing that does actually matter when you're actually working with audio, beyond fidelity..

Latency!

It would be awesome if all reviews involving ADC, DAC, or just digital processing included latency figures. Because when you're dealing with real time audio as a performer, latency is critical. It doesn't matter how good your gear sounds if there is a 50ms delay somewhere in the signal chain between the performance, and the artist hearing it.

Of course, I agree.

The earlier Lavry Gold AD converters latency was very high (around 8msec at 44.1KHz). The high latency offered advantage, time to do a lot of processing resulting in great audio conversion. But at the expanse of latency. These AD’s are used by mastering and other applications that don’t require low latency.

But the same Gold AD’s could not be used in situation such as over-dubbing. A singer listens to a recorded track and it’s voice delayed by AD, computer buffers, DA.

So the newer Gold AD (Savitr) is designed to have very low latency, including a special low latency mode. The latency is in the product specifications.

Savitr AD converter Latency:


Best Quality Latency: 471uS at 44.1KHz, 249uS at 96KHz, 130uS at 192KHz

Low latency mode: 153uS at 44.1KHz, 103uS at 96KHz, 86uS at 192KHz

Note: Both modes offer linear phase response (to retain the audio waveform shape).

​

 

[drumphil]

Using USB/ASIO on one side and a fixed hardware analysis on the other will give you these variations. ASIO or not - you are starting a software playback, and the OS will make sure that there is no fixed starting point in time for it (Windows/Mac are no real-time OS). ASIO used as both I/O can take care of that as record and playback are always started simultaneously, but here you remove the record part.

Whenever it is about timing you should use the digital output of the AP - that one is fully synced to the AP's digital and analog input, and thus allows to measure hardware latency in an I/O loop - as confirmed in your measurements.

@MC_RME

I would be grateful if you could provide some clarity on something else we've been discussing.

Is latency (not factoring in starting point, which as you have explained is variable) fixed with RME audio interfaces? If I take a RME USB audio interface, will I see the same round trip latency at a given buffer size, no matter which windows computer I use it with? Between mac systems? PC vs mac?

 

[drumphil]

The earlier Lavry Gold AD converters latency was very high (around 8msec at 44.1KHz).

Well, that should put to rest any notions of converter latency being too small to be consequential. It absolutely can be consequential for real time systems.

Have you made, or do you know of other converters with higher latency than that?

I appreciate that people like yourself are willing to provide this information for the high end products you make. It would be awesome if people starting out, with limited budgets, were able to make more informed decisions purchasing more affordable products. I think that is a great value that Audio Science Review can provide, this kind of information that manufacturers either haven't bothered to provide, or choose not to.

Of course, I still dream of a world where the work done by Amir and others wouldn't be necessary. Where I could pick up the box any audio interface came in, and read off the headphone output impedance. Where I could just read the specs for any ADAT converter rack, to know the ADC and DAC latency.

Meanwhile back in reality land, this site has greately reduced the amount of guessing and hunting around for information that needs to be done.