Does DSP latency matter?

[KEW]

Many subs have DSP's which delay the signal. I know an AVR with Audyssey corrects for this, but for those of us wanting to use a 2 channel receiver for music, is this something that just doesn't matter much?
There are many reviews of subs with DSP, but they never seem to mention timing as an issue??!!
Thanks!

 

[Steve81]

Many subs have DSP's which delay the signal. I know an AVR with Audyssey corrects for this, but for those of us wanting to use a 2 channel receiver for music, is this something that just doesn't matter much?

My guess is that the delay probably wouldn't make a world of difference given the human ear's poor performance at the frequencies in question. In Audyssey, I see maybe a foot or two added for the DSP in my PC12-NSD; not a lot to write home about. Given that in a simple 2 channel system you usually can't correct for distance at all, raw placement could have just as much (or more) impact.

Of course, the thought that comes to mind is that such an issue could be bypassed by utilizing the XO and high passed outputs on such a subwoofer.

 

[Gordonj]

Many subs have DSP's which delay the signal. I know an AVR with Audyssey corrects for this, but for those of us wanting to use a 2 channel receiver for music, is this something that just doesn't matter much?
There are many reviews of subs with DSP, but they never seem to mention timing as an issue??!!
Thanks!

Actually delaying the LF system is actually desirable in many cases. The delay of the sub allows your main speakers and the subs to form a cohesive wave front and arrive at the listener at relatively the same time. This "tweak" to the subs will in turn clean up the LF in your system. That is generally why you might find a DSP that has delay for a sub.

Gordon

 

[Gordonj]

Many subs have DSP's which delay the signal. I know an AVR with Audyssey corrects for this, but for those of us wanting to use a 2 channel receiver for music, is this something that just doesn't matter much?
There are many reviews of subs with DSP, but they never seem to mention timing as an issue??!!
Thanks!

Oops, one other thing.

Are you asking about DSP latency or the need for sub delay. Two very different things. Latency is an effect that is caused by the audio signal passing through the DSP device and the "delay time" from the input stage to the output stage. That latency is also dependent on the number of processing items you introduce to the signal path. Normally you do not "hear" this latency until you get to much larger systems where the latency of each device has added enough signal "delay" to be noticeable from an audio stand point.

Gordon

 

[TLS Guy]

Many subs have DSP's which delay the signal. I know an AVR with Audyssey corrects for this, but for those of us wanting to use a 2 channel receiver for music, is this something that just doesn't matter much?
There are many reviews of subs with DSP, but they never seem to mention timing as an issue??!!
Thanks!

Yes, latency is a big issue in digital systems. Current thinking is that latency needs to be as low as 30 to 40 samples, and there are now processors than can achieve those sort of latency specs. They are in the high end pro systems now.

In terms of sub latency, and timing the speakers in a system, this is something that Audyssey is very good at. Setting delays, by measured distance, is very rough at best. Setting levels, and delays, is in my view the principal benefit of Audyssey. I'm not sold at all on its ability to correct frequency response errors from loudspeaker systems, as the manner in which the response aberrations are produced by speakers are not amenable to simple frequency response correction.

 

[KEW]

Thanks for the responses.
So many of the subs now have DSP (Emotiva, SVS, for example) and there is no mention in reviews etc. about latency issues. While Audyssey matches any latency issues, these subs are presumably designed for use in an system that does not involve Audyssey (or equivalent).

I am just thinking out loud here, so please feel free to correct or dispute if you believe (or know) differently:
I suspect the 1-2 ft. at the speed of sound is probably challenging the limits of human perception (especially in these low frequencies, while the harmonics of the impact is still carried by the higher frequencies so will be in time with the main music).

I do have a specific application I am considering. It is using old school analog system with a SVS AS-EQ1 to EQ bass (the AS-EQ1 is Audyssey MultiEQ XT32 for subs only). I called SVS about this type of set up. The Tech Support rep said locating the subs 8 ft closer than the mains would compensate for it, but he was actually running that same type of set-up with the subs positioned next to the mains and hadn't noticed any difference. I'm sure he was telling the truth, but I also think it is reasonably likely that he might notice a difference if he had a corrected setup to A/B against.

For my case, I do not think I'll have trouble locating the subs 8 ft closer (though that might not allow for ideal sub placement).

 

[Gordonj]

Yes, latency is a big issue in digital systems. Current thinking is that latency needs to be as low as 30 to 40 samples, and there are now processors than can achieve those sort of latency specs. They are in the high end pro systems now.

In terms of sub latency, and timing the speakers in a system, this is something that Audyssey is very good at. Setting delays, by measured distance, is very rough at best. Setting levels, and delays, is in my view the principal benefit of Audyssey. I'm not sold at all on its ability to correct frequency response errors from loudspeaker systems, as the manner in which the response aberrations are produced by speakers are not amenable to simple frequency response correction.

I will add that the leaps in tech for DSP latency has already made its way down to the consumer level. Most of the DSP systems utilized for HiFi is much less demanding and processor heavy then those found in the pro level gear. Typically each frame of delay adds N(samples)*1/T(seconds of delay) to the system. Therefore, a delay (latency) of 16 samples at 48kHz equals 16/48,000= 333.3 micro-seconds. Not much time there. The older systems (back in the day) could add up to 10, 15, or even 20 milliseconds + of latency (delay) to the signal. That amount of delay wrecked audio signals and created real issues in the design of performance audio systems.

I agree that I do not believe in letting a computer equalize my system. Not so much due to the speaker capabilities but more from a room response stand point. A mic and a computer are "dumb" in the sense that they do not truly know the difference between a room modal and a speaker response. Therefore, if left unchecked, a notch filter in a computer can destroy the system response with you not even knowing what has occured. Having said that, most people are not familar with how to tune a system properly or even have the equipment to perform the task, so better that the computer do it for you.....

Gordon

 

[Gordonj]

Thanks for the responses.
So many of the subs now have DSP (Emotiva, SVS, for example) and there is no mention in reviews etc. about latency issues. While Audyssey matches any latency issues, these subs are presumably designed for use in an system that does not involve Audyssey (or equivalent).

I am just thinking out loud here, so please feel free to correct or dispute if you believe (or know) differently:
I suspect the 1-2 ft. at the speed of sound is probably challenging the limits of human perception (especially in these low frequencies, while the harmonics of the impact is still carried by the higher frequencies so will be in time with the main music).

I do have a specific application I am considering. It is using old school analog system with a SVS AS-EQ1 to EQ bass (the AS-EQ1 is Audyssey MultiEQ XT32 for subs only). I called SVS about this type of set up. The Tech Support rep said locating the subs 8 ft closer than the mains would compensate for it, but he was actually running that same type of set-up with the subs positioned next to the mains and hadn't noticed any difference. I'm sure he was telling the truth, but I also think it is reasonably likely that he might notice a difference if he had a corrected setup to A/B against.

For my case, I do not think I'll have trouble locating the subs 8 ft closer (though that might not allow for ideal sub placement).

Actually if you moved your subs 8' in front of the mains the subs would be "ahead" of your mains in time. Subs come with DSP's that are capable of delay so that they can be "moved" in time behind the mains. This delay (moving the subs back in time behind the mains) allow you to align the mains and the sub to be one wave front. And actually you can hear 1' or 2' of time when things are not aligned properly. It makes a very noticeable difference in sound quality (just 1ms sometimes which is just less then a foot).

Gordon

 

[Steve81]

I am just thinking out loud here, so please feel free to correct or dispute if you believe (or know) differently: I suspect the 1-2 ft. at the speed of sound is probably challenging the limits of human perception (especially in these low frequencies, while the harmonics of the impact is still carried by the higher frequencies so will be in time with the main music).

Yes. Speed of sound is ~1116 feet per second. Divided by 80Hz, and you're looking at a wavelength of about 14 feet. Applying what I know of group delay to the situation (ideal to keep it under 1 to 1.5 cycles) I would think as long as the delay of the sub + any delay of the DSP were kept under that threshold, you would be alright. Even 8' of delay could potentially be alright, depending on the sub, especially if you plan on crossing lower than 80Hz (given that the wavelength goes up as frequency goes down).

 

[KEW]

Actually if you moved your subs 8' in front of the mains the subs would be "ahead" of your mains in time.

To make a clarification, the subs DSP (typically in the form of a PEQ) is applied to the signal at the sub and only applies to the sub. Thus the sub's signal lags behind the main signal. In the case of the AS-EQ1, it is applied between the pre's sub out and the subs, giving the same net effect of lagging the sub's signal.

As Steve81 mentioned, you can send the main signal through the HPF of the sub (if the sub offers this feature) to introduce the same lag in the mains, but if you want subs at the back of your room, you might be talking 40-70 feet (round trip) for low-level signal. I would be very worried about picking up interference with that long of a run of standard RCA (unbalanced) cables.

 

[KEW]

And actually you can hear 1' or 2' of time when things are not aligned properly.
Gordon

I appreciate your input very much. Do you have a reference or link for this? No offense, but I would love to move this statement from hearsay to fact. I am also interested in seeing if frequency was included in analysis - I perceive a pretty substantial difference between my sensitivities in mid and high frequencies vs low frequencies.

 

[KEW]

Yes. Speed of sound is ~1116 feet per second. Divided by 80Hz, and you're looking at a wavelength of about 14 feet. Applying what I know of group delay to the situation (ideal to keep it under 1 to 1.5 cycles) I would think as long as the delay of the sub + any delay of the DSP were kept under that threshold, you would be alright. Even 8' of delay could potentially be alright, depending on the sub, especially if you plan on crossing lower than 80Hz (given that the wavelength goes up as frequency goes down).

Interesting concept that the number of cycles would have a greater correlation to our perception rather than the duration of the delay! Big difference between the duration of 1 cycle at 15KHz vs 30Hz (like 500 times quicker)!

 

[Steve81]

Interesting concept that the number of cycles would have a greater correlation to our perception rather than the duration of the delay! Big difference between the duration of 1 cycle at 15KHz vs 30Hz (like 500 times quicker)!

Indeed. If you look at some of Josh Ricci's subwoofer measurements, you can get a good visual of the delay.

Group delay of a PB12-NSD
http://www.data-bass.com/images/measurements/71/C pb12nsd group delay.jpg

and a spectrogram
http://www.data-bass.com/images/measurements/71/D pb12nsd spectrogram.jpg

You can see how as you approach 20Hz, the signal is delayed considerably as the tail on the spectrogram curves up.

Also worth noting, I believe based on a quick Q&A with Chris K. that Audyssey MultEQ can fix that as well. It's a marvelous tool.

 

[Gordonj]

I appreciate your input very much. Do you have a reference or link for this? No offense, but I would love to move this statement from hearsay to fact. I am also interested in seeing if frequency was included in analysis - I perceive a pretty substantial difference between my sensitivities in mid and high frequencies vs low frequencies.

No offense taken at all, the question keeps me on my toes for sure. In making these statements I have application experience to my aid. I have been a pro level systems designer/consultant/engineer for some 25+ years and have tuned too many sound system, performance systems, studio systems, theater systems, etc too remember. The comments that I make, and the opinions, are based on that vast experience in a career that i have been blessed to have been a part of and on top of that I am a LF nut and have done much experimenting with subs and time alignment.

Having said that here is one article that will take you deeper into the sub alignment world -

Live Sound: Phase Alignment Between Subwoofers & Mid-High Cabinets The Series: Part II - Pro Sound Web

Also, if you are truly interested in diving deeper into the tuning and alignment world the Sound System Engineering - Book by Don and Carolyn Davis could be considered one of the audio design bibles out there.

Gordon

 

[Gordonj]

To make a clarification, the subs DSP (typically in the form of a PEQ) is applied to the signal at the sub and only applies to the sub. Thus the sub's signal lags behind the main signal. In the case of the AS-EQ1, it is applied between the pre's sub out and the subs, giving the same net effect of lagging the sub's signal.

Understood perfectly. However, the latency (that you are referring to) would not be noticeable based on an earlier post to you questions that you are asking. There are home subs now that are adding to the DSP section the ability to add DELAY. Again, LATENCY is a much different item then DELAY. Latency is induced by the processing efforts of the DSP and, in this case, are very low in sample rates. In some cases you would want to add DELAY which would be a deliberate addition of time to the sub signal to align the two acoustic wave fronts.

Gordon

 

[Gordonj]

Indeed. If you look at some of Josh Ricci's subwoofer measurements, you can get a good visual of the delay.

Group delay of a PB12-NSD
http://www.data-bass.com/images/measurements/71/C pb12nsd group delay.jpg

and a spectrogram
http://www.data-bass.com/images/measurements/71/D pb12nsd spectrogram.jpg

You can see how as you approach 20Hz, the signal is delayed considerably as the tail on the spectrogram curves up.

Also worth noting, I believe based on a quick Q&A with Chris K. that Audyssey MultEQ can fix that as well. It's a marvelous tool.

According to the posted graphs that is some 20ms to 30ms of delay?? If that is indeed correct that is a substantial amount of delay! If it is truly ms then you are talking about roughly some 18' to 29' of physical time delay!!!!

Maybe Im just reading it wrong.

Gordon

 

[Steve81]

According to the posted graphs that is some 20ms to 30ms of delay?? If that is indeed correct that is a substantial amount of delay! If it is truly ms then you are talking about roughly some 18' to 29' of physical time delay!!!!

Maybe Im just reading it wrong.

Gordon

The measured delay ranges from ~10ms to over 100ms at 20Hz. The red and blue lines indicate the threshold of audibility, with the black line being actual measured. As you can see from that graph, it takes a considerable amount of delay to reach what Mr. Ricci indicates is the threshold of audibility. Based on that data, I wouldn't expect 1-2 feet to do much of anything. I could see 8 feet being problematic if crossed high enough and if the sub was already near the threshold of audibility.

 

[Steve81]

According to the posted graphs that is some 20ms to 30ms of delay?? If that is indeed correct that is a substantial amount of delay! If it is truly ms then you are talking about roughly some 18' to 29' of physical time delay!!!!

PS: Given the above speed of sound (per The Google!) of 1116 feet per second, 1116 feet = 1000ms, so 1ms = 1.116 feet. So 20ms should be 22.3' and 30ms should be ~33.5'. Correct me if I'm wrong.

 

[Gordonj]

PS: Given the above speed of sound (per The Google!) of 1116 feet per second, 1116 feet = 1000ms, so 1ms = 1.116 feet. So 20ms should be 22.3' and 30ms should be ~33.5'. Correct me if I'm wrong.

You are close in your speed calls (actual speed at STP is 1087.42 fps). Your calls for delay are fine. And yes even 1millisecond (hence a foot) can be noticed and can make a noticeable difference. That is why in large speaker system boxes the highs can be delayed just 1.345 milliseconds. It makes a difference in the wave front! And you can see the arrivals on a graph and they are frequency dependent.

Gordon

 

[Gordonj]

The measured delay ranges from ~10ms to over 100ms at 20Hz. The red and blue lines indicate the threshold of audibility, with the black line being actual measured. As you can see from that graph, it takes a considerable amount of delay to reach what Mr. Ricci indicates is the threshold of audibility. Based on that data, I wouldn't expect 1-2 feet to do much of anything. I could see 8 feet being problematic if crossed high enough and if the sub was already near the threshold of audibility.

And can you see a problem with those calcs?

Gordon