• 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!

Drawbacks of software and hardware equalization and how to counter them?

anphex

Addicted to Fun and Learning
Forum Donor
Joined
May 14, 2021
Messages
662
Likes
870
Location
Berlin, Germany
A few days ago I used REW to measure my speakers in different rooms and let it calculate the filters for a equalization below 250hz with +-3db linearity tolarency. When I put those into Equalizer APO it spat out a heavily corrected curve since my rooms and listening position are far from optimal. I had to set a gain reduction in APO of about 12db!

Even though the sound was noticably improved it made me think. What price am I paying with such a tremendous correction?

My first thoughts are:
- reduction of the maximum volume of the system due to heavily boosted frequencies that may clip way earlier
- more strain on the amps
- phase shifts, though I am not sure if there is a way to adjust between minimum and linear phase in APO
- struggle with the neighboors if the sub bass was boosted greatly when sitting in a bass hole

The CPU processing power of 1 core @ about 1% I consider negligible.

These are for software equalization. Further for hardware EQs/DSP:
- more noise and distortion in the chain since most gadgets perform sub par compared to high quality DACs
- more complicated signal chain and more points of failure
- increased cost


The feasible counters for some of the cases that came to my mind:
- the more ways a speaker has, the more tolerant it is to eq since frequencies are further distributed between chassis
- speakers with low distortion can handle heavy equalizations more gracefully
- for every 3db gain twice the power of a "normal" amp should be considered


Of course there are also benefits on both sites. Software equalization makes an external devices unnecessary but it's only active on PC. The other way round hardware gadgets enable to EQ for all connected devices on the chain but comes with all drawbacks listed above.

Do I overthink some drawbacks here? Or are there even more one should be aware of?
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,227
Likes
17,807
Location
Netherlands
My first thoughts are:
- reduction of the maximum volume of the system due to heavily boosted frequencies that may clip way earlier

Boost less and make sure you partially solve the issues with room optimization. Deep valleys are not solvable by DSP anyway.

- more strain on the amps

Get a bigger amp, or see above. Also, don't forget that strain on the drivers will be much worse than on the amps in most cases.

- phase shifts, though I am not sure if there is a way to adjust between minimum and linear phase in APO

If you are worried about that use FIR.

- struggle with the neighboors if the sub bass was boosted greatly when sitting in a bass hole

Move! But really: bass holes cannot be DSP-ed away. Change the location of speakers, sub or listener or apply better room treatment.

These are for software equalization. Further for hardware EQs/DSP:
- more noise and distortion in the chain since most gadgets perform sub par compared to high quality DACs

That has nothing to do with the software. Just use adequate interfaces or DAC's, and you will have no noise issues.

- more complicated signal chain and more points of failure
- increased cost

Possibly. A decent PC + DAC8 can still be cheaper than a professional hardware DSP though.
 

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,823
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
I highly recommend and suggest you to try multichannel multi-driver (multi-way) multi-amplifier system with DAC8PRO and software EQ-DSP on PC; each of the sub-woofers, woofers, midrange squawkers, tweetters and super-tweeters should be driven by dedicated amplifier (eliminating all the LC inductors, capacitors, resistors and attenuators) with of course separate independent relative gain....
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
My first thoughts are:
- reduction of the maximum volume of the system due to heavily boosted frequencies that may clip way earlier
- more strain on the amps
- phase shifts, though I am not sure if there is a way to adjust between minimum and linear phase in APO
- struggle with the neighboors if the sub bass was boosted greatly when sitting in a bass hole

As mentioned by @voodooless you need to be careful when applying boosts. It is also very important to understand how those boosts will be implemented in your DSP and where you volume control is. You will have a lot more flexibility if you do volume control upstream of your DSP because you can raise the volume when source material is mastered at a low level AND avoid clipping the DSP when you are listening to higher level material that requires a much lower volume setting.

Just my two cents but I would expect smoother phase response after correction. If you are getting something else it likely means that you do not fully understand your measurements / corrections. To me the only phase shifts that are worth correcting with FIR filters are inherently non-minimum phase behavior like that created from using a crossover.

These are for software equalization. Further for hardware EQs/DSP:
- more noise and distortion in the chain since most gadgets perform sub par compared to high quality DACs
- more complicated signal chain and more points of failure
- increased cost

In general I agree that hardware DSP is more expensive and does not perform as well. That being said I think hardware DSP is way simpler to implement and in general has a much lower learning curve. My intuition also says that hardware DSP is much more reliable but both have been reliable in my experience. The biggest disadvantage for software EQ is that it is much more difficult to setup especially if you are using it tie together different input and output devices.

Of course there are also benefits on both sites. Software equalization makes an external devices unnecessary but it's only active on PC. The other way round hardware gadgets enable to EQ for all connected devices on the chain but comes with all drawbacks listed above.

It is very easy to implement software DSP such that it is applied to many different input sources as long as you have the appropriate input functionality. All of my systems now use RPis running software DSP but have conventional input functionality (AES, TOSLINK, SPDIF, etc).

Michael
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
2,920
Likes
3,834
reduction of the maximum volume of the system due to heavily boosted frequencies that may clip way earlier

...- struggle with the neighboors if the sub bass was boosted greatly when sitting in a bass hole
Cutting usually isn't a problem. Boosting is where you quickly get into trouble with clipping your amp and/or driving your speakers into distortion (especially in the bass). With digital EQ you can easily push your DAC into clipping, especially if you're not using software volume-control.

You generally can't fix a "bass hole" with EQ because it can take an almost infinite amount of power and infinitely large woofers. Changing the speaker/woofer location may help and multiple woofers can help by "spreading around' the soundwaves. The magic of bass traps is that they trap the bass that would otherwise be reflected and that smooths-out the dips and the peaks. (I'm no telling you to use acoustic treatment, and I don't have any, but it's a good place to start if you want a "perfect room".)

phase shifts
Phase shifts aren't a big deal unless the phase is shifted relative to something. For example, if you flip the polarity of one speaker by reversing the connections the left & right will be out-of-phase. That causes bass cancelation and a weird "spacey" sound. But if you flip the connections to both speakers everything sounds normal again.
 

digitalfrost

Major Contributor
Joined
Jul 22, 2018
Messages
1,521
Likes
3,086
Location
Palatinate, Germany
Wheter your loose those 12dB in the digital or analog domain does not really matter. If you had the same EQ in the analog domain, you'd also need to take care not to overdrive the inputs of the power amplifiers and you would need a similar gain reduction. Given how cheap software EQ is and the possibilities it provides, I see no reason for hardware EQ.

Those 12dB are gone either way. Analog/digital is just two ways to look at the same thing.
 
OP
anphex

anphex

Addicted to Fun and Learning
Forum Donor
Joined
May 14, 2021
Messages
662
Likes
870
Location
Berlin, Germany
I tried so many EQ adjustments and in the end the best result was just setting about 4 PK filters manually. REW auto adjustment ist either very aggersive or not doing anything at all. At least I haven't found found yet an option where you can "limit" the adjustments. It either irons out the peaks and dips or keeps them.

But damn, I hate my small wood house. It's probably not even wave cancellation, the walls literally eat any multiple frequencies from 63 hz on.

Edit: When adjusting speakers in APO with separate EQ for every speaker make sure you add the "Channel: L" or "Channel: R" in the text file. If you forget it - like me of course - the result will be awful.
 
Last edited:

bigjacko

Addicted to Fun and Learning
Joined
Sep 18, 2019
Messages
721
Likes
359
I tried so many EQ adjustments and in the end the best result was just setting about 4 PK filters manually. REW auto adjustment ist either very aggersive or not doing anything at all. At least I haven't found found yet an option where you can "limit" the adjustments. It either irons out the peaks and dips or keeps them.

But damn, I hate my small wood house. It's probably not even wave cancellation, the walls literally eat any multiple frequencies from 63 hz on.

Edit: When adjusting speakers in APO with separate EQ for every speaker make sure you add the "Channel: L" or "Channel: R" in the text file. If you forget it - like me of course - the result will be awful.
Did you eq based on your in room measurement ? It is better to eq based on klippel measurement or anechoic measurement so you don't correct some high frequency reflections in the room.
 
OP
anphex

anphex

Addicted to Fun and Learning
Forum Donor
Joined
May 14, 2021
Messages
662
Likes
870
Location
Berlin, Germany
Did you eq based on your in room measurement ? It is better to eq based on klippel measurement or anechoic measurement so you don't correct some high frequency reflections in the room.

No only corrected a few boosts and wide dips in the bass range below 200hz.
 

Plcamp

Addicted to Fun and Learning
Joined
Jul 6, 2020
Messages
860
Likes
1,315
Location
Ottawa
I used minidsp2x4hd extensively on open baffles to do things like try out differing crossover topologies, eq approaches etc.

Having done that I’d say…

- Avoid high q filters and don’t try to fix minor response perturbations. Exception might be cone breakup notches.
- Make sure you understand the speaker’s own in-situation resonances, reflections, diffraction so you don’t try to eq the impossible.
- No net dsp gain at any frequency, except of course between channels to match levels.
- Minidsp low frequency filters can be problematic. I avoid them except for low q shelving.
- it adds noise in signal chain. Minor but there.
- it sounds poorer on analog in than digital in.
- I suspect and others claim it is noticeably inferior vs modern DAC performance.
- It’s flexibility is addictive.
 
OP
anphex

anphex

Addicted to Fun and Learning
Forum Donor
Joined
May 14, 2021
Messages
662
Likes
870
Location
Berlin, Germany
I used minidsp2x4hd extensively on open baffles to do things like try out differing crossover topologies, eq approaches etc.
- Avoid high q filters and don’t try to fix minor response perturbations. Exception might be cone breakup notches.

This was one of my problems yesterday. My left speaker has a huge high q dip at exact 200hz of about -24db. I gave it a gain high Q of +9db but I am still not sure. Should I just leave it? I'll just post my current settings later. Maybe this thread can also be used for EQ consulting.
 

Plcamp

Addicted to Fun and Learning
Joined
Jul 6, 2020
Messages
860
Likes
1,315
Location
Ottawa
This was one of my problems yesterday. My left speaker has a huge high q dip at exact 200hz of about -24db. I gave it a gain high Q of +9db but I am still not sure. Should I just leave it? I'll just post my current settings later. Maybe this thread can also be used for EQ consulting.

I would want to know why that dip was there before I tried to dsp it. Seems it’s a room effect given you see it only on one speaker.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
This was one of my problems yesterday. My left speaker has a huge high q dip at exact 200hz of about -24db. I gave it a gain high Q of +9db but I am still not sure. Should I just leave it? .

dips that low are usually infinitely low (complete cancellation of the wave by boundary interaction or axial room modes), there is nothing that can be done except treating the room effectively to fix it.
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
In what world is a 1000$ DAC + God knows how much PC is cheaper than an Audio interface with PEQ or any hardware MiniDSP-esque device.

This one. :)

If you actually need 8 channels of output the dac8 pro + RPi compares very nicely to the miniDSP offerings (especially given their current product lineup with no nanoDIGI and no OpenDRC-DA8). The closest miniDSP product is the 4x10HD which is $590 once you add the phoenix to XLR output adapter. This has no display, no trigger output, no FIR filtering, limited PEQ, limited delay, worse analog performance and can only operate in a 2 in, 8 out configuration. Okto + RPi has none of those limitations although it does have more limited input functionality (AES + USB)

I agree that you can use an audio interface + RPi in a similar manner to the Okto + RPi at a lower price point but you still have issues to deal with. Most reasonably priced audio interfaces do not have an IR receiver so remote volume control is not an option. You can of course use software volume control but adding a separate IR receiver and display requires more work / expense. I have not seen any audio interfaces with trigger outputs but you can add a Bobwire DAT1 if they have TOSLINK output although this again requires more work / expense.

Michael
 
OP
anphex

anphex

Addicted to Fun and Learning
Forum Donor
Joined
May 14, 2021
Messages
662
Likes
870
Location
Berlin, Germany
Here are my manual corrections. I hovered over the predicted line to make it glow since REW seems to pick almost identical colors for base and predicted lines :/.
As per @abdo123 s suggestion I removed the useless peak at 200hz and now it's only manually "fixed" below 123 hz. If anyone sees a way to make improvements here I'd happy if you'd share it.
 

Attachments

  • NVIDIA_Share_rWYpI7TSga.png
    NVIDIA_Share_rWYpI7TSga.png
    209.3 KB · Views: 126
  • roomeqwizard_e4w45JaWIL.png
    roomeqwizard_e4w45JaWIL.png
    156.8 KB · Views: 125

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Here are my manual corrections. I hovered over the predicted line to make it glow since REW seems to pick almost identical colors for base and predicted lines :/.
As per @abdo123 s suggestion I removed the useless peak at 200hz and now it's only manually "fixed" below 123 hz. If anyone sees a way to make improvements here I'd happy if you'd share it.

your response is actually quite healthy without EQ, but it seems that your speaker is not anechoically flat since there are a lot of serious dips above 2KHz. and there is not much I would recommend to fix that.
 
OP
anphex

anphex

Addicted to Fun and Learning
Forum Donor
Joined
May 14, 2021
Messages
662
Likes
870
Location
Berlin, Germany
your response is actually quite healthy without EQ, but it seems that your speaker is not anechoically flat since there are a lot of serious dips above 2KHz. and there is not much I would recommend to fix that.

That's nice to hear. The speakers are +-3db 23hz-24khz by manufacturer spec. I could imagine that either the mic, my handling of the mic, the glass table, the tv right between the speakers could have produced that wigglyness in the highs. Or it's just D'Appolito-combfilter when you go off axis. When I sit on my sweet spot on the couch I am usually still about 20cm below the tweeter. I could check for the latter later.

Edit: I added the picture with var smoothing for both speakers to improve visibility. That's raw measurement of both individual speakers.
 

Attachments

  • roomeqwizard_TlIf0UW1aZ.png
    roomeqwizard_TlIf0UW1aZ.png
    852.4 KB · Views: 93

Xombul

Active Member
Forum Donor
Joined
Apr 1, 2019
Messages
123
Likes
128
Location
France
I believe that a cheap raspberry 4 or a similar nano computer is powerful enough for room correction, no need to invest in an expensive hardware solution. To apply correction filters I have found Camilla DSP very flexible and quite easy to handle.
 

Matias

Master Contributor
Forum Donor
Joined
Jan 1, 2019
Messages
5,031
Likes
10,803
Location
São Paulo, Brazil
Here are my manual corrections. I hovered over the predicted line to make it glow since REW seems to pick almost identical colors for base and predicted lines :/.
As per @abdo123 s suggestion I removed the useless peak at 200hz and now it's only manually "fixed" below 123 hz. If anyone sees a way to make improvements here I'd happy if you'd share it.
Personally I prefer to use Psychoacoustic smoothing to ignore the small peaks and dips and focus on the large deviations, and then let the EQ run on top of it. It creates a less aggressive correction.
Then export the filters as wave files and use convolution software to apply them (even Equalizer APO supports this) so that phase shifts are not an issue.
 
Top Bottom