Sonarworks Reference 4 Headphone Edition Review

Life in the Age of DSP

I recently received my pre-order Audeze Mobius, a collection of crossfeed software such as NX Waves, Goodhertz Canopener Studio and the subject of this review: Sonarworks Reference4 software.

To put it mildly it’s had me thinking about the place of DSP and correction in the land of audiophiledom.

In the recording studios and venues I work in, correction, while not ubiquitous, is quite common in the professional audio realm. However, just because a recording studio does it, does not necessarily mean it is right for every context.

Whether you are of the pleasant coloration or total transparency camp, audiophiles are evaluating more than just frequency response. Adding signal processing or additional components always changes that, so why would we want to add more DSP to our signal chain? Especially in a headphone listening system, where we enjoy much higher levels of detail than speakers, free of room interactions, crossovers and dozens of gain stages and volume controls.

Sonarworks has become a fairly well known company in the professional sound world recently for its speaker and headphone-calibration software. Their speaker calibration especially is one of the more interesting efforts out there, with correction for time and phase issues and multiple points of room interaction. Sonarworks has also participated quite actively in the head-fi and super best audio friends communities, sharing and requesting information on their headphone measuring process quite openly. These folks seem to pay attention to detail and listen to their user base, a big plus in my book.

Sonarworks currently has two products of interest to the headphone enthusiast – their audiophile oriented ‘true-fi’ software and the more expensive and professionally targeted Reference 4 software. I am reviewing the Reference 4 software here, which compared to the true-fi version includes a higher-degree of specific functionality for professionals, including a measurement microphone and two-channel tuning software.

Differences between the pro and consumer-oriented version are a target curve and filter-type options, and an upcharge of $10. I would buy the Reference Headphone product every time over the true-fi as the functionality it offers is more than worth the small cost. I would only recommend true-fi to those for whom $10 is a lot of money, or who consider themselves less than expert at navigating computers and software – in which case you might have taken a wrong turn on the Googles.

In the box of the full Reference 4 package Sonarworks includes a generic-measurement microphone, but otherwise the entire product is downloadable software. Setup is relatively easy, just enter your serial code and select your output device. The Sonarworks ‘systemwide’ apps shows up as a playback device which all of the audio from your computer runs through. On launch the application gives you a choice of some 232 headphone models to choose from. Select the model you have, and it loads a calibration file based on Sonarworks measurements and averaging of several pairs of that headphone.

Once the calibration file is loaded, you are shown a ‘post-correction’ curve, with the option to also display the correction-target curve, phase shift, or assumed pre-correction curve. You can enable or disable the correction to A/B the effect, and also add bass boost, or several tilt filters. There is a generic EQ-tilt filter which can go from 1-6db per octave tilts either moving bass up and treble down or vice versa. There is also a 1974 B&K -3db per octave speaker curve, and an ‘x-curve’ which is described as a motion-picture standard curve. There is also an option to adjust output, dry/wet control for adjusting the effect to taste, mono and automatic-gain compensating ‘safe headroom’ controls, and a zero-latency or linear-phase filter option.

I spent time looking at every single correction curve in Sonarworks’ software, comparing it both to Innerfidelity measurements as well as some of my own and the community’s measurements of many popular headphones. The Sonarworks measurements seem to have the same basic shape as Innerfidelity measurements, though I do see noticeable differences on several headphones, for example, the Sonarworks measure responses seemed to show less low-bass energy and seemed to be more sensitive in the 5-6khz range. Measurements in the treble are always fraught; unit variance, measurement technique and equipment, there are a host of difficulties in getting reliable, repeatable measurements.

I will note however, that most of the headphones measured seem to resemble the shape of Innerfidelity measurements using the Harman Curve correction. In listening, I found most of the corrections ended up with something fairly close to the Harman Curve, although usually a good bit bassier in the very lowest octaves, and occasionally a little brighter than I would like in the presence region. Treble was surprisingly well-behaved and I had very few issues there – somewhat the opposite of what I would have expected looking at the measurements.

Sonarworks SIA Smerla iela 3 Riga, LV-1006 Latvia

buckchester's picture

Great review. I tried Sonarworks with a pair of Hifiman HE-400i and I found it improved the sound quite a bit. However, I felt like it boosted the bass a little too much and it added a subtle sort of “shouty” quality to the sound, which I believe lies somewhere in mid range between 1khz and 5khz.

I’ve since bought a MiniDSP EARS and MiniDSP HA-DSP, which by the way, I’m able to get even better sound with. With these, I have been able to significantly improve the sound quality of the following headphones: Audio Technica M50x, Hifiman HE-400i, and Focal Elex.

After measuring the 400i headphones with the EARS, the bass response I measured is quite close to most other measurements of out there (quite close to those SBAF, actually). Sonarworks measurements of these headphones show a very steep roll-off in the bass starting around 100hz, however, most other measurements, including mine, show a much narrower roll-off in this region. When I measure these headphones with the Sonarworks curve engaged, it shows a massive (~+12db above neutral) bump in the 40hz region. As a result of their measurements, I believe they have over-compensated in this area with these headphones. I wonder if those that they measured had worn out earn pads, which can apparently cause the bass to roll-off more steeply due to a flimsier fit. I found their bass measurements of the Audio Technica M50x and the Focal Clear Professional (closest they have to the Elex) seem closer to what I get with the EARS.

I cannot hear a difference between the filters.

Grover's picture

Hi buckchester

I actually have a mini DSP ears and HA-DSP unit right now and am working on a review of it. It's a pretty sweet little setup for the money. I'm trying to do a small survey of DSP-related produccts right now - nxWAves, miniDSP, sonarworks, etc. This is a big area audiophiles typically aren't fans of, but I think we're starting to see some very compelling products here.

I also find the sonarworks plots to have far too much bass in them, especially on more open headphones. Not sure why that is, as my experience measuring is that treble is usually more finicky than bass - in this case Sonarworks seems to have excellent treble, but occasionally odd bass and midrange measurements. I think miniDSP's idea of only cutting and not boosting and using a unity gain correction is really on the right path - I do wish you could use separate converters and amps with it though. It's also quite a bit trickier than sonarworks obviously, which does all the hard stuff for you.



Skycyclepilot's picture

I used Equalizer APO, a peak filter in that program, and a Voxengo graphic equalizer VST (hosted by Equalizer APO) to match my DT 1990s to the FR curves published by and DIY Audio Heaven with excellent results. All Beyer headphones have a sharp peak at about 8 KHz, thus the peak filter. The 1990s are overly warm, thus the equalizer. If you're willing to invest the time, this method is free.

echoplex's picture

Thank you for the review. It's unclear to me from looking at the screenshot of the software how you would get Sonarworks 4 to equalize the headphones to reproduce the Harmon Curve. Clearly, Sonarwoarks could have added a target curve for this, but it appears they did not. While I have no doubt that equalizing to a flat response will improve the sound of inexpensive headphones, the idea of taking a headphone that is already well designed to match the Harmon Curve - and then forcing it to flat response - makes no sense technically if you want you mixes to sound the same in headphones as they do in a well treated room (which is what audio professionals would be doing). This would not be a flaw in the Sonarworks design *unless* they are really reproducing the Harmon curve when they picture "flat" response on the graph.

I have never used the mini-DSP boxes, but you should also look into the software measurement application and plugins in IK Multimedia's Arc 2.5 system (cones with a different kind of design for a measurement mic). ARC predates Sonarworks and uses the Audyssey MultEQ® XT32 correction software found in some surround sound units.

Then at a more sophisticated level of acoustical correction (and cost), is the Dirac room correction software, and Trinnov correction (only available in hardware) which many pro audio people use depending if they want a software or hardware solution.

Slawomir's picture

In my opinion equalization of headphones should be based on measurement without(with subtracted) Harman Curve because human ear create resonances causing bump in 3khz region and headphones actually don't cause this phenomenon. Engineers from Sonarworks could and should equalize headphones to target without Harman Curve if they're using dummy head with ear canal.