miniDSP Tide16 - Holy Grail with 16 Channel Atmos/DTS:X, high SINAD

[thcdru2k]

Exactly! The only "room correction" above the Schroder frequency should be "broad, low-Q, tone control equalisation" as Toole puts it. But this does not mean that full range DSP is not possible. It is of course possible, but that is speaker correction and not "room correction". What that means is, take a quasi-anechoic measurement of speaker or driver, and use that as basis for correction. A lot of those software packages don't give you the flexibility of taking a separate measurement and performing a limited correction of that.

I've never understood the concept let's take any system and make them all fit the same curve . Calibration should be system/room specific.

 

[TurtlePaul]

Exactly! The only "room correction" above the Schroder frequency should be "broad, low-Q, tone control equalisation" as Toole puts it. But this does not mean that full range DSP is not possible. It is of course possible, but that is speaker correction and not "room correction". What that means is, take a quasi-anechoic measurement of speaker or driver, and use that as basis for correction. A lot of those software packages don't give you the flexibility of taking a separate measurement and performing a limited correction of that.

If you are creating your own DSP software (like Dirac) there is more you can do than this. I agree that a raw frequency response sweep of an unknown speaker in an unknown room will not generate a usable basis for an EQ. However, with more sophisticated software you can do variable gating to tighten the gate after first impulse to reduce reflections at higher frequencies (based on assumptions for how big resonable size rooms are or info about distance from speaker to listening position). You can also correct the impulse response - which does nothing audible really but people love the idea of time alignment.

 

[Keith_W]

I've never understood the concept let's take any system and make them all fit the same curve . Calibration should be system/room specific.

I'm not sure what you are getting at here. With all types of calibration, there is a calibration target. For example, if you bought a TV, you would try to calibrate it towards colour accuracy. It is the same with bass correction, the target is a smooth bass response. It's not the same with upper frequencies however, which is the point I was making earlier. If you look at the actual correction being done, it is system/room specific. Even specific down to where you took the measurement from. But the target is the same?

 

[thcdru2k]

I'm not sure what you are getting at here. With all types of calibration, there is a calibration target. For example, if you bought a TV, you would try to calibrate it towards colour accuracy. It is the same with bass correction, the target is a smooth bass response. It's not the same with upper frequencies however, which is the point I was making earlier. If you look at the actual correction being done, it is system/room specific. Even specific down to where you took the measurement from. But the target is the same?

The target can be the same, but you can’t make every system hit it the same way. Above the bass range, a speaker’s inherent response and directivity dominate, and sometimes it makes sense to preserve its natural tonality. If I had money for an expensive speaker, I would want to hear what it actually sounds like, not force it to trace a curve.

In the bass, many dips are geometric cancellations. EQ can’t fix those, and boosting into them just wastes headroom and adds distortion. So the target might be fixed, but the usable correction is system and room dependent.

I'm sure this is up for discussion but just what I think.

 

[Bjorn]

This is off-topic here, but it's a common misunderstanding to think that gating the impulse works well for correction. It's more complex than that. That's been possible with for instance Audiolense for a long time, but the result is still generally very unatural sound due to what Heyser called phase distortion.

I hope miniDSP can comment if Tide 16 will have FIR for crossovers and FIR for manual correction.

 

[peng]

I hope miniDSP can comment if Tide 16 will have FIR for crossovers and FIR for manual correction.

I hope so too, probability of that seems low, given that the Tide 16 will come with Dirac Live so it is always "activated". In the case of their simpler devices such as the Flex 2X4HD, it seems that as others mentioned, if Dirac Live is activated (paid option), you lose the FIR (manual). Not really sure why miniDSP took such approach, but I guess ultimately it is a limitation of the DSP.

 

[Ultrasonic]

In the bass, many dips are geometric cancellations. EQ can’t fix those, and boosting into them just wastes headroom and adds distortion.

This is true for complete cancellations but in practice many are not, meaning that some boosting can be beneficial. Stereo Dirac Live does this. I have no experience with DLBC or ART but I'd expect these may do something similar.

 

[Tell]

I remember a time when MiniDSP made affordable stuff. I still love my MiniDSP 2x4HD, but if they'd update it to take multichannel PCM, added 2-4 RCA outputs (I have no need for XLR), variable loudness and a web interface it would be even better. But instead they make this monstrosity that at least I will never ever buy.
Ah well, great for those that need and can afford this!

 

[Sokel]

This is true to complete cancellations but in practice many are not, meaning that some boosting can be beneficial. Stereo Dirac Live does this. I have not experience with DLBC or ART but I'd expect these may do something similar.

DIRAC was forced to do that after a lot of push back, mainly by those who like to watch this straight declining line at the charts and know nothing else about it.
Dips should NOT be filled by DSP, even in doubt.

 

[Ultrasonic]

DIRAC was forced to do that after a lot of push back, mainly by those who like to watch this straight declining line at the charts and know nothing else about it.
Dips should NOT be filled by DSP, even in doubt.

Probably a discussion for another thread so I'll simply say that I disagree.

 

[Sokel]

Probably a discussion for another thread

Yes, we already have lots of them with top experts explaining why not.
OT, out.

 

[Keith_W]

Dips should NOT be filled by DSP, even in doubt.

It would be more accurate to say that dips should not be filled by boosting the amplitude of the equalisation. Whether a dip can be filled with DSP depends on the cause of the dip. If it is caused by a sound interacting with its reflection, then this hole is infinitely deep and requires infinite SPL to fill it. But if it is caused by two speakers interacting with each other, then the phase of the narrow band can be adjusted in one speaker to remove or reduce the dip in the sum. This is why the eBook contains a procedure to help you diagnose the cause of a dip.

 

[peng]

It would be more accurate to say that dips should not be filled by boosting the amplitude of the equalisation. Whether a dip can be filled with DSP depends on the cause of the dip. If it is caused by a sound interacting with its reflection, then this hole is infinitely deep and requires infinite SPL to fill it. But if it is caused by two speakers interacting with each other, then the phase of the narrow band can be adjusted in one speaker to remove or reduce the dip in the sum. This is why the eBook contains a procedure to help you diagnose the cause of a dip.

I have a stupid question that may be you can help me understand. If a dip is caused by room mode, I understand boosting with something like a PEQ filter will not work, at least not effective, but I am always curious about the following:

- let's say a +6 dB gain PEQ will not budge such a dip of -6 dB at 90 Hz
- yet, if I increase the MV by 6 dB, everything would go up 6 dB including that dip right?

How is it different, between a boost at 90 Hz only, and increase the master volume by the same 6 dB at 90 Hz only?

When I had the last Denon AVR, I had used the MultEQ editor app to flatten the bass range just for fun, and was able to flatten room mode dips by lowering the level from 20-20 kHz by say 3 dB, and then was able to boost dips including those I thought were room mode dips. Doing it that way, I was able to get 20-150 Hz to within +/- 1 dB or even less 1/12th smoothing. Thank you in advance.

 

[EngineerNate]

I would guess that it’s because the dip won’t be a fixed 6db at every output level.

 

[voodooless]

yet, if I increase the MV by 6 dB, everything would go up 6 dB including that dip right?

The depth of the dip very much depends on the smoothing used when displaying. Due to this, is appera as if the dip moves with the gain, but in reality, it does not.

Also consider that not all dips are full nulls. But I would not try to boost them anyway, unless they are like 3db down without much smoothing.

 

[peng]

I would guess that it’s because the dip won’t be a fixed 6db at every output level.

That's my rationale at the time and that's why I used that approach and it worked! But, in theory, increasing the volume dial position, while it does increase the output level across the board, it still is the same as increasing the gain. So I understand by lowering say 6 dB across the board, the localized dips such as the example one at 90 Hz will get filled, the dip would reoccur if the volume is return to the original (higher) position when listening, yet that's not the case as I could raise the volume to the level I needed, and the dip would still be gone, or whatever improvements achieved.

 

[TurtlePaul]

How is it different, between a boost at 90 Hz only, and increase the master volume by the same 6 dB at 90 Hz only?

It will fill the dip and this works and it sounds better.

The real rationale for not filling dips is to avoid blowing speakers and amps. 10 dB dip sounds like half volume but requires 10x power to fill. Set the peak boost to somewhere in the 3-6 dB range and fill away. If you have 10-20 dB dips, then speaker positioning is the only real solution.

 

[EngineerNate]

Yeah that’s a more detailed version of what I was trying to say. At some point the dip is so big something in your chain lets go or gives up.

 

[peng]

The depth of the dip very much depends on the smoothing used when displaying. Due to this, is appera as if the dip moves with the gain, but in reality, it does not.

That's not the case though as you can see in the graph below. It was not the best achieved, as this one would be more like +/-1.5 dB 20-130 Hz, but you can still see that big improvements on just about every dips within that range, and that would not be improved much if I did not lower the levels across the board first.

No smoothing used in this case:

 

[peng]

Yeah that’s a more detailed version of what I was trying to say. At some point the dip is so big something in your chain lets go or gives up.

For clarity, I did not over boost to force an effect, let me used an example below, that's how I did it initially but had improved on it later.
The adjustments made were just simple arithmetic, plus or minus, as seen from the as found curve post Audyssey.

Steps take to shape the target curve via MultEQ Editor and Ratbuddyssey:

Enter the values to a spreadsheet table as show below, or you can do it manually but spreadsheets are much quicker.

New target SPL	68
Freq on ref curve	SPL on ref curve	Distance from target	Adjustments
13	70.84	2.84	-2.84
15	73.06	5.06	-5.06
16	71.95	3.95	-3.95
19	69.56	1.56	-1.56
22.96	68.75	0.75	-0.75
27.61	72.86	4.86	-4.86
33.6	73	5	-5
35.7	73	5	-5
39	70.48	2.48	-2.48
41	71.55	3.55	-3.55
45	70.47	2.47	-2.47
49.7	64.23	-3.77	3.77
51.6	65.99	-2.01	2.01

Edit: Sorry to get way off topic, so I will stop here, and may be our moderator could move the posts to an appropriate thread of just delete them. Thanks to all those responded.