DSP/Room Correction and High Sample Rate Audio

[thrand1]

Hi,

 

I am not very well technically versed in this area, so I apologize in advance for any misinformation or other inaccuracies in my following post

 

I was reading an article on room EQ produced by Dr. David Rich over at Secrets of Home Theater and High Fidelity. The article focused primarily on Anthem's ARC system and how it works and was interesting for people like me that have some understanding of room EQ, but not on a very technical level.

 

One interesting item of note at the conclusion of the article was that, in most consumer level home theater receivers using Audyssey EQ (and Anthem's MRX home theater receivers), with Audyssey or ARC room equalization engaged, signals with 96kHz sample rates were "downsampled" to 48kHz for Room EQ purposes. I would guess this is probably due to hardware limitations/resource demands placed on the receiver. It was noted in the article that Anthem's higher-end processor, the D2V, can perform room EQ up to (and including) 96kHz sample rates.

 

With this in mind, in our world of very high resolution downloads plus DSD, does utilizing room EQ minimize the "differences" one would hear between different sample rates/DSD since those signals may be reduced to a common sample rate?

 

Additionally, although the article doesn't test additional rates, are sample rates that are higher multiples of 44.1kHz (88.2, 176.4) downsampled and possibly converted to 48kHz? This likely is not as clean of a conversion, either.

 

With discussions of products like Dirac in this forum, is the limitation of performing room EQ functions at higher sample rates (i.e. >=96kHz) largely due to physical resource limitations, or something else? For example, is even having a dedicated PC with an i5/i7 processor not enough horsepower to perform room EQ without downsampling "higher resolution" (i.e. DSD, 24/192 and above) files?

 

Even with the downsampling to 48kHz for Audyssey, would there be any benefit from feeding it a higher sample rate file? For example, does sending a 192kHz signal into the DSP for room EQ purposes provide any benefit over say a 96kHz file?

 

This may be an open-ended discussion, but I just want to learn more in this area. I personally have a very sub-optimal room, and Audyssey EQ has made a significant, positive difference in the sound in my room, particularly when it comes to subwoofer/main speaker integration. While I have been branching into computer audio and learning more about it and planning future hardware purchases, I almost feel more reluctant to purchase hardware for even "higher resolution" playback (I have only 24/96 files/playback capability at this time) if I am still going to be downsampling the file, whether it's 24/96, 24/192, DSD, to a lower sample rate.

 

If you have any experiences, anecdotes, or other insights to share, please feel free to post them here

[S.o.k.]

Hi. I Think there is a Big difference in terms using receiver ment for film in a HT. setup or using a dsp. ment only for stereo! Even if you do have the option of using roomcorrection in high ress over 96/24. i am not shure it is the best solution, if the roomcorrection is ment to work over the hole bandwith! I belive that roomtreatment is still the best way above 100hz. Under 100hz. roomcorrection very effektive and in my experince, the best way to achieve control. Hope this is usefull for you. Kind regards.

[sdolezalek]

The limitation in every chain is its weakest link and frequently it is hard to find that weak spot. Spending a lot of money on a his res 24/192 front end makes far less sense when somewhere else in the chain everything is run through 16/48 processing.

 

I had those same concerns in my own system. My pre/pro specifies that it has two 24-bit/192kHz digital-to-analog converters for each of its 8 audio output channels; but if you read the fine print on the room and bass management processing it says: Four 32-bit floating-point DSP engines that manage this processing "at sample rates up to 96kHz with 24-bit resolution to retain top performance from all input sources."

 

So I don't get much benefit, and may in fact be adding one unnecessary up and down conversion if I was to use 192kHz source material AND run it through any of the sound processing steps. Fortunately, the unit has the ability to defeat all tone controls, so I can run 192 kHz material unaltered.

 

In your case, I'm not sure whether you can bypass the 48 kHz processing.

[sshd]

With this in mind, in our world of very high resolution downloads plus DSD, does utilizing room EQ minimize the "differences" one would hear between different sample rates/DSD since those signals may be reduced to a common sample rate?

 

What difference?

 

- Very few people can tell the difference if the resampling has been done properly.

 

- Very few people can hear above 24 kHz - the limit imposed by 48 kHz sample rate.

 

DSP is not the usual audiophile quest for perfection. It is more about finding a great compromise, working with what you have.

 

DSP is based on math and therory related to how normal humans hear and perceive sound. If you like what you hear then the theory is obviously correct.

 

If you wish to improve the sound, try other DSP software solutions like Dirac, Audiolense, DRC-sourceforge and Acourate. You can listen before you buy.

[thrand1]

Hi. I Think there is a Big difference in terms using receiver ment for film in a HT. setup or using a dsp. ment only for stereo! Even if you do have the option of using roomcorrection in high ress over 96/24. i am not shure it is the best solution, if the roomcorrection is ment to work over the hole bandwith! I belive that roomtreatment is still the best way above 100hz. Under 100hz. roomcorrection very effektive and in my experince, the best way to achieve control. Hope this is usefull for you. Kind regards.

 

I agree with you especially in room correction under 100Hz. Due to the irregular shape of my room and the fact it is "open" to other areas, when I conducted a frequency response sweep of the room I had variations of +/-8dB from 20Hz to 120Hz! Using the Audyssey room equalization plus a 3band parametric equalizer on my subwoofer brought me to a tighter +/-3dB response in the same frequency range.

 

As you mentioned, treatment of the actual room is often most effective. You can't/should not try to boost "dips" in the frequency response, those items can only be "treated" by optimizing the listening position or position of speakers, etc.

 

The limitation in every chain is its weakest link and frequently it is hard to find that weak spot. Spending a lot of money on a his res 24/192 front end makes far less sense when somewhere else in the chain everything is run through 16/48 processing.

 

I had those same concerns in my own system. My pre/pro specifies that it has two 24-bit/192kHz digital-to-analog converters for each of its 8 audio output channels; but if you read the fine print on the room and bass management processing it says: Four 32-bit floating-point DSP engines that manage this processing "at sample rates up to 96kHz with 24-bit resolution to retain top performance from all input sources."

 

So I don't get much benefit, and may in fact be adding one unnecessary up and down conversion if I was to use 192kHz source material AND run it through any of the sound processing steps. Fortunately, the unit has the ability to defeat all tone controls, so I can run 192 kHz material unaltered.

 

In your case, I'm not sure whether you can bypass the 48 kHz processing.

 

You can bypass the 48kHz processing by putting the receiver into a "Pure" or "Direct" mode which bypasses the Audyssey processing entirely. In the article in my original post, the author demonstrated this by passing an 18kHz tone and a 30kHz tone through the processor. With Audyssey engaged, the 18kHz tone passed, but not the 30kHz tone (as it is above the 24kHz analog frequency limit imposed by the 48kHz sampling rate). With Audyssey disengaged, both tones were visible on the plot.

 

What difference?

 

- Very few people can tell the difference if the resampling has been done properly.

 

- Very few people can hear above 24 kHz - the limit imposed by 48 kHz sample rate.

 

DSP is not the usual audiophile quest for perfection. It is more about finding a great compromise, working with what you have.

 

DSP is based on math and therory related to how normal humans hear and perceive sound. If you like what you hear then the theory is obviously correct.

 

If you wish to improve the sound, try other DSP software solutions like Dirac, Audiolense, DRC-sourceforge and Acourate. You can listen before you buy.

 

I guess I should have rephrased that- if everything you pass into a receiver is being resampled to 48kHz for room EQ processing, then would the potential differences (audible or otherwise) between different sample rates be minimized or eliminated? No matter if it's a 96kHz file, 192kHz file, DSD file, or DXD/352.8kHz file, all of it will be "converted" to that 48kHz sample rate, making me think that any potential for noticing differences (if you the individual can even pick them up!) would be pretty much eliminated.

 

Thanks for the feedback everyone, some interesting things to read about and learn!

[TomScrut]

Are you sure that Audyssey is limited to 48KHz? On the thread in the link below it would appear that Chris Kyriakakis from Audyssey is being quoted as saying the DSP only works under 30KHz, but it doesnt downsample it just passes everything else through above that frequency. Unfortunately I couldn't find the original post on Ask Audyssey.

 

"Official" Audyssey thread (FAQ in post #51779) - Page 1619

 

That doesn't mean that the receiver manufacturer isn't down-sampling prior to the DSP though.....

 

Either way, I use MultiEQ XT32 Pro on my AV8801 and think it sounds a hell of a lot better than with it switched off!

[Kal Rubinson]

Are you sure that Audyssey is limited to 48KHz? On the thread in the link below it would appear that Chris Kyriakakis from Audyssey is being quoted as saying the DSP only works under 30KHz, but it doesnt downsample it just passes everything else through above that frequency. Unfortunately I couldn't find the original post on Ask Audyssey.

 

"Official" Audyssey thread (FAQ in post #51779) - Page 1619

That doesn't mean that the receiver manufacturer isn't down-sampling prior to the DSP though.....

 

Either way, I use MultiEQ XT32 Pro on my AV8801 and think it sounds a hell of a lot better than with it switched off!

I do not know what quote you are referring to but Audyssey, itself, is not limited to 24/48. However, the limitation on the resolution at which it can work is determined by the DSP resources of the hardware in which it runs and, unsurprisingly, manufacturers are somewhat stingy about that since the fixed amount of DSP installed must run many tasks simultaneously (e.g., decoding of HD codecs). Chris's statements attempt to justify the common 24/48 limitation.

 

Either way, as you imply, one can decide for himself, with or without. I tend to vote with you.

[TomScrut]

Sorry I thought the URL was tagged with the appropriate post. I was just arguing against the first post in this thread that suggests that audyssey downsamples when it doesn't.

[thrand1]

Kal, Tom,

 

What term/terms would you use to describe what Dr. Rich (from the first post) found when passing an 18kHz tone and then a 30kHz tone through his Marantz unit...clearly with Audyssey engaged nothing above 23-24kHz shows up on the resulting graphs, while with Audyssey disabled the frequency response/tones present changes.

 

Would the presence/absence of tones not suggest some sort of downsampling/other change in the signal is occurring? Maybe I am just confused by the terminology. Is Audyssey taking an input signal, "changing" it somehow to 48kHz for DSP purposes, then "changing" it again back to its original incoming rate?

 

I am not trying to say that the DAC in the unit is incapable of decoding anything "above" 48kHz, obviously if you feed signals up to/including 192kHz in the Marantz they would be decoded/you would get sound...my question is more with Audyssey in the loop, is anything above 48kHz incoming sampling rate being "changed" to that 48kHz number to perform Audyssey operations? Again, looking for a better term to replace "changed" to describe the process...

[TomScrut]

It could be the marantz downsampling before inputting to audysseys DSP. I haven't read the original article you posted, I will do when I get a minute. I am disappointed this affect is the case on a marantz as I have one too!

[TomScrut]

I have read the article now. Coincidentally it's an 8801 I have! I will do some digging online tonight to see if it's a limitation of the technology or the marantz, and they didn't mention whether it had been pro calibrated just a price comparison. It could be that a pro calibrated one can do more as the DSP chip may not be the limitation, it could be, as they say themselves about the anthem, that the pc contributes to the processing power when working the filters out and therefore the marantz's internal processor for the filters could be the weak link. I am intending getting a calibrated mic before long so could do some tests.

 

Either way, pro install on Audyssey is loads better than the standard mic setup.

[TomScrut]

Ok, so I have found out a few things, including a conversation I participated in and forgot about!

 

Denon AVR-5805: Audyssey and System Set Up | Audioholics

 

This is the article about Audyssey and its capabilities etc but not the one quoted in the thread I posted earlier. It does mention 96KHz support though.

 

Marantz Flagship AV8801 Processor | Page 7 | AVForums

 

This is the discussion about this issue on the Marantz, and it would appear that it is not enough DSP processing power to allow it to remain at 96KHz. I do wonder if this is only an issue with XT32 or if the lesser variants also struggle for processing? I say that as the resolution of the filters on XT32 are a lot higher than XT or standard MultiEQ. One thing I am disappointed in is that on stereo material playing through two speakers it should be able to process the higher sample rates, as its meant to be able to use 13 speakers altogether!

[Miska]

At general level, many AVRs run internal DSP processing at one particular fixed sampling rate. It is usually ether 48k, 96k or 192k. What I've seen most myself is 96k rate being used, but I think some more modern ones are supporting 192k.

 

For any other input sampling rates, they run the signals through sample rate conversion.

[Kal Rubinson]

Kal, Tom,

 

What term/terms would you use to describe what Dr. Rich (from the first post) found when passing an 18kHz tone and then a 30kHz tone through his Marantz unit...clearly with Audyssey engaged nothing above 23-24kHz shows up on the resulting graphs, while with Audyssey disabled the frequency response/tones present changes.

 

Would the presence/absence of tones not suggest some sort of downsampling/other change in the signal is occurring? Maybe I am just confused by the terminology. Is Audyssey taking an input signal, "changing" it somehow to 48kHz for DSP purposes, then "changing" it again back to its original incoming rate?

 

I am not trying to say that the DAC in the unit is incapable of decoding anything "above" 48kHz, obviously if you feed signals up to/including 192kHz in the Marantz they would be decoded/you would get sound...my question is more with Audyssey in the loop, is anything above 48kHz incoming sampling rate being "changed" to that 48kHz number to perform Audyssey operations? Again, looking for a better term to replace "changed" to describe the process...

 

I am not sure what you are asking. What seems to be happening is that, when Audyssey is invoked and you have an input with higher bit rate/depth, the signals are downsampled to 24/48 for processing. 24/48 and lower bit rate/depth signals are unaffected What happens afterward is not defined but once the 30kHz signal is lost, it is lost.

[Kal Rubinson]

I have read the article now. Coincidentally it's an 8801 I have! I will do some digging online tonight to see if it's a limitation of the technology or the marantz, and they didn't mention whether it had been pro calibrated just a price comparison. It could be that a pro calibrated one can do more .................................

Pro calibration makes no difference.