Sonarworks worth it for headphones?

[DLS79]

To be clear I'm not talking about for audiophile usage, but for critical listening/mixing/mastering.

I know there are a lot of potential issues when it comes to measuring headphones, and unit to unit variances etc, but setting that aside for a moment, do you think Sonarworks is worth it at a conceptual level?

 

[PaulD]

I have tried it, and I would say no, it's not worth it. As a generalisation I found the corrections in the bottom end to be ok or worthwhile, but corrections in the higher frequencies to be appalling to the point where I could not use it.

I might be ok if it allowed selective correction over a limited frequency range, but it does not.

I think Sonarworks needs to do a lot more work on target reference curves and also to allow correction only over a defined frequency range.

By comparison I have found Waves NX to be far superior (with the clip-on dongle). I know it is less of a frequency and response correction system and more of a spatialisation system, but it makes listening to headphones far closer to listening in the studio.

 

[mechapreneur]

At a conceptual level, I think Sonarworks is onto something. Headphones have peaks and dips in reproduction. All speakers have them. Room correction is the same concept for speakers and seems to make speakers and rooms sound better together. I don't have a problem with the concept.

What I question is the value for money.

I've been toying around this weekend with AutoEQ and a parametric EQ on my MacBook and wow, what a difference it makes. I've got a new pair of Fostex T50RP MK3 headphones and they were sounding a bit bright and fatiguing in the high end while having reasonable bass. With the settings from AutoEQ, that bright edge is gone and the bass is even more impressive. I'm sold.

 

[solderdude]

I have been EQ-ing headphones long before the digital variants existed.
What I found is that while you can get great measurement results compensating exactly what is measured on a specific rig and get a nice flat line on that particular setup with the exact same headphone on rig position it will be off when measured on any other rig as well as on any other head.
It's a bit like applying room correction measured at the couch and then listen at the listening chair (other position).

Sure, there are problem areas that are improved of course. Large dips, peaks and roll-off is compensated for after all.

In the end the result also depends on the target curve one aims for which as is also known does not work for everyone equally well and is listening volume (habit) depending as well.

What I found was that 'exact' correction on a specific test rig is not the way to go and one can be many dB's off on another test rig or actual ears.
When my EQ is made (analog, not really a digital DSP guy myself) I just look at my measurements and those of others and look for common factors.
Make an EQ that comes close. Have a listen and sometimes make small changes (or big ones) and measure and listen again.
This ends up with somewhat different EQ's than Sonarworks, Oratory or Jaakko (or other corrections) on the web but works well for me.
So my p.o.v. is to only correct what has been seen to be an issue and to a target curve that works for me.

For me EQ works wonders but one really cannot make a HD201 sound like a HD800 on all aspects but can get the tonal balance close enough.

 

[thewas]

Very true words, I tried to EQ my HD600 to a Focal Clear (whose sound signature I really like) based on the differences of measurement of both from different measurement setups (which were also all different) and none sounded like the Focal.

Also Sonarworks EQs works for me well only for some headphones and for others not.

I guess this is due to the fact that our ear geometry is different to a measurement ear and also (same with loudspeaker measurements) a measurement microphone just measures the total sound pressure almost omnidirectionally while our hearing apparatus weighs different directions differently. That's why I doubt that opposite to loudspeaker corrections, fully automatic headphone corrections based on current measurement techniques will be as good for now.

 

[DDF]

I've been toying around this weekend with AutoEQ and a parametric EQ on my MacBook and wow, what a difference it makes

I can also highly recommend Toneboosters Morphit, for UAPP, for an extra $5 as an in app purchase. It comes with eq settings for a boatload of headphones, and can toggle numerous target settings including the Harman curve. I compared its eq for my SRH840 with the parametric eq settings at AutoEQ (using Toneboosters parametric eq), and for this isolated case the outcome was noticeably more natural, pleasant and accurate sounding. No brainer $5 purchase, it has most of my other headphones in there as well. Only drawback is that these tonebooster eqs do use up allot more battery on my tablet.

 

[b1daly]

If you have access to a decent EQ, I suggest EQ’ing by ear to where you think it sounds good. Ideally you want a fully parametric EQ.

There are two broad conceptual areas: The overall tonal balance, highs to mids to lows. This is usually pretty obvious and is subject to personal preference anyway.

The other area are “correcting” unpleasant “resonances.

This is tricky, but one way to do it if you have a true parametric EQ is to set a very narrow band. Put a significant boost on it, 6-10dB. “Sweep” the frequency to while listening. It will sound pretty gnarly, but when you hit a resonant peak, it will stand out. When you find this frequency, use a narrow band EQ and pull that frequency down, by ear, usually just a couple dB.

The inverse, correcting resonant dips is harder, but it can be done. In this case you are looking for areas where your annoying peak that you sweep isn’t quite as annoying as the adjacent frequencies.

When you get it right, compare to the dry signal. It should now sound like an “annoying” EQ is being put on the mix when you are actually removing it

It helps to have a handful of tracks you like and can get familiar with.

One of the challenges with correcting speaker response is the body of mixes out their have a wide range of spectrums. Trying to EQ on a super bright mix will lead to cutting treble too much. You want to get a sense of the “average”.

Another thing that can make this hard is that EQ changes the overall level. So adding any frequency boost will make the mix louder. We tend to prefer louder mixes in AB comparisons. To get around this, you can pull the volume all the way down with the EQ on, and then bring it up to listening level to see if you like the sound.

 

[Veri]

I personally use Sonarworks and love it to bits. I use it for a HD650 and a modded audio technica which I sent in to sonarworks for a calibration profile. I always have it on.

I use both just under 70% 'wetness' to retain just a little of the headphone character. 100% 'flat' sounds just not as good to me, which imo is not the best way to use this software. If you're using to a 'flat' calibrated headphone and don't like the sound, I suggest tweaking with the settings. Except for phase, I leave it to low latency. Can't hear any difference anyway..

 

[DLS79]

If you have access to a decent EQ, I suggest EQ’ing by ear to where you think it sounds good. Ideally you want a fully parametric EQ.

Just as an aside, it doesn't matter to me personally if it sounds good or better. I'm more interested in the mixing and mastering aspects where a lot of times what you want doesn't sound that great.

This is tricky, but one way to do it if you have a true parametric EQ is to set a very narrow band. Put a significant boost on it, 6-10dB. “Sweep” the frequency to while listening. It will sound pretty gnarly, but when you hit a resonant peak, it will stand out. When you find this frequency, use a narrow band EQ and pull that frequency down, by ear, usually just a couple dB.

I do this a lot when trying to find the sibilant frequencies in vocals.

Another thing that can make this hard is that EQ changes the overall level. So adding any frequency boost will make the mix louder. We tend to prefer louder mixes in AB comparisons. To get around this, you can pull the volume all the way down with the EQ on, and then bring it up to listening level to see if you like the sound.

A good trick for this is to loudness normalize the two tracks you are comparing. Since I'me mostly concerned with audio for video I normalize to EBU R128.

 

[DLS79]

I've been toying around this weekend with AutoEQ and a parametric EQ on my MacBook and wow, what a difference it makes.

That's a helpful site, I'll bookmark it and spend a little time looking through it.

 

[DivineCurrent]

Sonarworks got me into EQing headphones, and most of the time their EQ profiles sound good to me. However, there is still too much variance for the methods they use. According to their compensation curve, the HD650 and 600 have a 2-3 dB dip at 2 kHz, which I do not hear at all. If anything, it's elevated in that area using sine sweeps, to my ears. They do have an individualized EQ calibration service where you send them your own headphones, and they measure that set to their curve with an accuracy of +-1 dB. That may be worth it assuming your ear pads don't drastically change shape over time. Whether the service is worth $149 though is up to you.

 

[svenz]

I believe the variance is in the headphones themselves. I asked Sonarworks about this directly and they sent me various measurements showing as much as 6db variance in some frequencies even in expensive TOTL headphones. It was quite eye opening. Also ear pads deforming over time can have a big impact on the FR. I still think Sonarworks is worth it, it substantially improves all my headphones.

 

[robertozappala]

I can also highly recommend Toneboosters Morphit, for UAPP, for an extra $5 as an in app purchase. It comes with eq settings for a boatload of headphones, and can toggle numerous target settings including the Harman curve. I compared its eq for my SRH840 with the parametric eq settings at AutoEQ (using Toneboosters parametric eq), and for this isolated case the outcome was noticeably more natural, pleasant and accurate sounding. No brainer $5 purchase, it has most of my other headphones in there as well. Only drawback is that these tonebooster eqs do use up allot more battery on my tablet.

Sonarworks plays with Roon and Audirvana, Morphit only with Audirvana

 

[DavidS]

I’ve been playing with Toneboosters with UAPP to play Qobuz track on several headphones. It sounds pretty impressive and in the same league as the improvements I get from my RME ADI-2 DAC - though I’m nowhere near having done any serious comparisons. Assuming you have an Android phone available it seems a cost-effective option.

Can anyone tell me how it works? Does it do a D/A conversion, apply the EQ and then do an A/D conversion back to feed the USB DAC? All on the fly?
David

 

[svenz]

I’ve been playing with Toneboosters with UAPP to play Qobuz track on several headphones. It sounds pretty impressive and in the same league as the improvements I get from my RME ADI-2 DAC - though I’m nowhere near having done any serious comparisons. Assuming you have an Android phone available it seems a cost-effective option.

Can anyone tell me how it works? Does it do a D/A conversion, apply the EQ and then do an A/D conversion back to feed the USB DAC? All on the fly?
David

It's purely digital until your DAC converts it to analogue.

 

[DavidS]

Ok, I clearly don’t understand how the format used for storing digital music allows you to apply eq. Can you point me at a reference or even explain. I realise there is structure to the form of the digital information but I don’t know how that structure relates to db uplifts at various points in the frequency range. Apologies if this is a badly-formed question. David

 

[svenz]

Ok, I clearly don’t understand how the format used for storing digital music allows you to apply eq. Can you point me at a reference or even explain. I realise there is structure to the form of the digital information but I don’t know how that structure relates to db uplifts at various points in the frequency range. Apologies if this is a badly-formed question. David

https://www.musicdsp.org/en/latest/_downloads/550af030275b020f8f85b422e699ce5f/EQ-Coefficients.pdf
https://www.musicdsp.org/en/latest/_downloads/3e1dc886e7849251d6747b194d482272/Audio-EQ-Cookbook.txt

This is where all digital EQ filter implementations come from, I believe. You'll need to understand signal processing at a formal level to derive and understand the formulas. But suffice to say, you can replicate an analogue EQ in the digital domain.

 

[DavidS]

That is fantastically helpful - thinks. it will stretch my maths degree to really understand but I get the basic idea. I appreciate you taking the trouble to respond and so quickly. David

 

[Scoox]

I can also highly recommend Toneboosters Morphit, for UAPP, for an extra $5 as an in app purchase. It comes with eq settings for a boatload of headphones, and can toggle numerous target settings including the Harman curve. I compared its eq for my SRH840 with the parametric eq settings at AutoEQ (using Toneboosters parametric eq), and for this isolated case the outcome was noticeably more natural, pleasant and accurate sounding. No brainer $5 purchase, it has most of my other headphones in there as well. Only drawback is that these tonebooster eqs do use up allot more battery on my tablet.

I'm testing demos of Morphit and Sonarworks and toggling between the two it's very obvious they are doing very different things to the audio, so one (or both!) must be "wrong".

 

[DDF]

I'm testing demos of Morphit and Sonarworks and toggling between the two it's very obvious they are doing very different things to the audio, so one (or both!) must be "wrong".

I developed an EQ method that noticeably improves my headphone enjoyment so sharing it below in the hope it helps others.

I measured Morphit (android) and Sonarworks (desktop) stock EQs and they did differ. Both were perceptually more neutral (my preference) than the free internet EQs on my 5 or 6 headphones (YMMV) so were worth exploring further.

But both had some big drawbacks that I wanted to overcome:
- Morphit allows target adjustment but limits it to a few fixed targets including Harman, "HiFi", "Studio" (less bass than Harman)
- Morphit EQs up to too high a frequency. Response above ~6 kHz especially depends on headphone seating and individual ear shape.
- Sonarworks supports a user defined max frequency for equalization
- Sonarworks doesn't work with Tidal or Quobuz for lossless streaming
- Sonarworks cost (99Euro vs Morphit's $10 in UAPP)
- Each is platform specific, meaning you need to buy more than one SW

I got around these issues:
- Bought Morphit (in UAPP) as I found it more neutral (stock) than Sonarworks and the <1/10 price doesn't hurt
- Measured the EQ shape I liked best using REW
- Reverse engineered that EQ using REW to generate PEQs for use in Toneboosters on android (comes with Morphit) and EAPO on desktop
- Critically I then tailor the PEQ settings to my preference (every head is different) and usually turn off the PEQs above 6kHz to avoid the aforementioned issues

If Eq'ing to a high frequency, you also need to know the PEQ filter type used by your equalizer (and sample rate) if you want accurate EQ. Digital filter "cramping" is where the PEQ filter "bell" shape gets increasingly asymmetrical as it approaches half the sample rate. Some equalizers correct for this differently (or not at all). This is one reason REW supports multiple equalizer settings.

"Generic" filters generated by REW match EAPO and don't have cramping correction. The EQ is still accurate and what you see with REW is what you get with EAPO.

Toneboosters heavily corrects cramping so I asked Toneboosters support to suggest an REW filter setting matching their design, but they understandably declined to answer (but were otherwise very helpful). I recommend PEQs be turned off above ~ 6 kHz if possible so it happens to work out that cramping isn't a major issue with Toneboosters using this method.