SoundID SR For Headphones: Measurement examples and implications for mixing

[abe_zydar]

Question: Is Sonarworks SoundID SR a superior method of determining FR in headphones? In the link below they have graphs of headphones measured using different hardware and propose that they've developed a better way of determining the desired audio output. I think this is important because of how music is mixed since many producers/mixers use SoundID for calibration.

https://www.sonarworks.com/blog/research/white-paper

Some background (and another question at the bottom) on why I'm wondering this....

Today, I was comparing equalization curves from Oratory1990, Crincale, and Sonarworks SoundID for my Audio Technica ATH-R70X headphones. I'm using SoundSource from Rogue Amoeba to do A/B tests and volume matching the output. Granted, SoundID is doing more processing on the signal than just EQ as it also tries to phase match the output based on their ATH-R70X wired average. I don't hear much difference with phase matching disabled though.

ASR, Crinacle and Oratory1990's measurements show a dip in the 4-5k region with a stronger dip at 10k (though not sure if that should be considered).

Here is SoundID's measurement, which is a bit confusing as their target is "flat", but that's a bit of a black box. For Sonarworks, "flat" means they've matched an equalization curve similar to studio monitors.

Here's some rough estimates on measured differences:

100hz -> SoundID: +3db from target, ASR: -1db, Oratory: +0.5db
3k -> SoundID: -3db from target, ASR: ~-3db, Oratory: 0db
5k -> SoundID: -3db from target, ASR: -7.5db, Oratory: -3.5db
7k -> SoundID: +3db, ASR: Close, Oratory: Close

Listening to the difference is striking in the mids and highs. As a music producer who is trying to find out a "reference", this seems a bit bewildering. So I guess I have another question for other producers/mixers on this forum. Do you mix using a Harman target for headphones or try to go for something "flat"?

 

[markanini]

Sonarworks uses a different artificial ear simulator than Harman reasearch so it's hard to make direct comparisons.

I personally seek out headphones that are Harman compliant in 1-5KHz and spend time listening to familiar songs and make an manual EQ until it sounds tonally correct.

But in the end I think headphone mixing is a tangent, the majority of decisions are made on speakers and a AB-ing with commercial reference tracks.

 

[andymcbain]

Here is SoundID's measurement, Listening to the difference is striking in the mids and highs. As a music producer who is trying to find out a "reference", this seems a bit bewildering. So I guess I have another question for other producers/mixers on this forum. Do you mix using a Harman target for headphones or try to go for something "flat"?

Very good question and one that I don’t have an answer for quite yet. As someone who’s done a lot of mixing in headphones over the years, I was interested to learn from ASR that the HD650 seem fairly compliant to the harman curve. When I bought the HD650 in the late 2000s it was (rather unscientifically) recommended as one of the models to get if you had any desire to mix in headphones - which was less common back then. What I consider to be my best mixes were done in those cans, despite a lot of guess work / trial and error for the low end. I shied away from headphone mixing in general when my tinnitus set in after a particularly rough night of DJing.

I’ve picked up the HD650 again as a mixing tool recently and remembered why I liked working on them so much. I do have the Sonarworks plugin, but the only time I seem to enable it (alongside Goodhertz Canopener) is when I’m mixing a vocal and balancing sibilance. For whatever reason, the combination of these two plugins just brings that sibilance into razor sharp focus for me. Unscientific, I know! But I’ll spend some time looking into why that may be.

Sonarworks also helps immensely with the HD650s low end - though my preference for mixing that with confidence is now the Slate VSX headphone + plugin. For context, I’m mixing and producing part time in a small terraced house with neighbours either side so large monitor speakers and copious bass trapping aren’t really an option

Reference tracks have always been useful for me - albeit tracks that are carefully selected. Though l can see why that method would be counter intuitive if we’re moving towards mixing with a more scientific approach.

And so ends my ramble

 

[abe_zydar]

@markanini @andymcbain Thank you both for taking the time to respond!

But in the end I think headphone mixing is a tangent, the majority of decisions are made on speakers and a AB-ing with commercial reference tracks.

I think you're right about this. Doing some further research I came across this on the iZotope website: https://www.izotope.com/en/learn/expert-mastering-engineers-on-headphones.html. Most mastering engineers use headphones in a limited fashion, but I wonder if that will change over time.

though my preference for mixing that with confidence is now the Slate VSX headphone + plugin

I always felt wary of the marketing tactics with Slate plugins, so I've been a bit skeptical of Slate VSX. Good to hear that perhaps they have something that works beyond the marketing hype. I'd be interested to hear your thoughts on why you prefer it to Sonarworks.

I also came across this lengthy comment by Oratory1990 on the differences between different headphone targets and why the Harman target seems to be the top preference. If the Harman target truly tries to capture the response of a flat studio monitor in a headphone, that seems to be at odds with what Sonarworks is saying.

https://www.reddit.com/r/headphones/comments/eetyx9/_/fbxbp9j
He ends with this (DF and FF mean Diffuse Field and Free Field)

The difference to DF and FF curves is that the room will slightly boost low frequencies due to reverberation, but high frequencies do not reverberate as much as they are more easily absorbed.
This comes much closer to what the artist and recording engineer heard in the studio, and what they based their judgement on in order to shape the sound of the music.
In other words: The Harman Target is basically the same sound that the artist and engineers heard when creating the music that we hear.

 

[andymcbain]

I always felt wary of the marketing tactics with Slate plugins, so I've been a bit skeptical of Slate VSX. Good to hear that perhaps they have something that works beyond the marketing hype. I'd be interested to hear your thoughts on why you prefer it to Sonarworks.

Oh the marketing is annoying as hell. Thankfully my regular mastering engineer was a beta tester for VSX and I trust his opinions - so I got to ignore all the hype. As a fellow Brit, this review from Russ Hughes of Pro Tools Expert basically nails it for me - https://www.pro-tools-expert.com/pr...late-audio-vsx-headphone-mixing-system-tested

There’s no question that VSX is a useful product. For me I mostly use it for getting the low end right in both production and mixing. The sub bass response is very very good - and when combined with whatever psychoacoustic stuff the software is a doing, there’s an almost tactile sense of bass from these headphones and the low end reacts with whatever environment is being modelled in a useful way - so I can dial in precise cuts and boosts. Sonarworks helps the HD650 sub bass considerably, but in comparison to VSX it feels a bit like painting a hallway through a letterbox.

£500 seems a little steep for the VSX package but I’d have to spend way more than that to upgrade my room and monitors to a similar standard.

What they don’t really give me is that mid range detail - so it’s my HD650s all the way for that. If I can get the foundations right in VSX though, everything falls into place a lot quicker and easier.

Thanks for all the information on the Harman curve - I’ll certainly spend some more time studying it. My gut feeling is well, my HD650s are relatively compliant, all my favourite mixes sound good in them and the tracks I attempt to mix using them come out ok though I won’t entirely rule out the Sonarworks plugin for the reasons I mentioned earlier.

 

[markanini]

I recommend manual corrective headphone EQ using familiar songs. If you are easily overwhelmed by choice, like me, limit yourself to an octave band EQ, 1dB or .5dB steps. Personally this has done more for me than either Sonarworks or Oratory1990 presets.

 

[Zensō]

@markanini @andymcbain Thank you both for taking the time to respond!
If the Harman target truly tries to capture the response of a flat studio monitor in a headphone, that seems to be at odds with what Sonarworks is saying.

Sonarworks uses the term “flat”, but they’re actually speaking in relation to their in-house target, which is similar to Harman’s but with a lower bass shelf.

 

[abe_zydar]

Ah excellent, that’s exactly what I was wondering

 

[Phorize]

Sonarworks uses the term “flat”, but they’re actually speaking in relation to their in-house target, which is similar to Harman’s but with a lower bass shelf.

View attachment 289967

I'm interested in your opinion as a producer on the additional gain that sonarworks put into the 2-3K region in their 'flat' curve. Last night I was using a HD560s to a/b between the sonarworks 'flat' profile, stock HD560s and oratory1990s profile. I really appreciated the more moderate mid bass gain in the sonarworks profile compared to Harman but found the 2-3k region pretty 'shouty' with some quite undesirable spatial effects that I need to consider more to put my finger on. I was quite surprised that this was considered 'flat' in professional circles-the effect is to take a pretty neutral headphone, significantly improve the bass but to make it otherwise far less neutral, at least to me.

 

[Zensō]

I'm interested in your opinion as a producer on the additional gain that sonarworks put into the 2-3K region in their 'flat' curve. Last night I was using a HD560s to a/b between the sonarworks 'flat' profile, stock HD560s and oratory1990s profile. I really appreciated the more moderate mid bass gain in the sonarworks profile compared to Harman but found the 2-3k region pretty 'shouty' with some quite undesirable spatial effects that I need to consider more to put my finger on. I was quite surprised that this was considered 'flat' in professional circles-the effect is to take a pretty neutral headphone, significantly improve the bass but to make it otherwise far less neutral, at least to me.

I’ve compared Sonarworks vs Oratory on a variety of headphones with inconsistent results. On one headphone SW sounds closer to my monitors, on another Oratory’s EQ sounds closer. I think this illustrates the problem of so many variables in the process including individual HRTF, leakage, different test fixtures and procedures, unit-to-unit variance, and so on. With all of these variables, what sounds “right” with a certain target, on your head, with your particular headphones, may be different than what sounds right to the next person. Any or all of these variables could easily swamp the differences between these targets. My approach is to use whatever correction most closely matches my headphones to my calibrated monitors. I know this is not a satisfying answer, but I do think it describes the reality of the situation.

 

[Zensō]

I'm interested in your opinion as a producer on the additional gain that sonarworks put into the 2-3K region in their 'flat' curve. Last night I was using a HD560s to a/b between the sonarworks 'flat' profile, stock HD560s and oratory1990s profile. I really appreciated the more moderate mid bass gain in the sonarworks profile compared to Harman but found the 2-3k region pretty 'shouty' with some quite undesirable spatial effects that I need to consider more to put my finger on. I was quite surprised that this was considered 'flat' in professional circles-the effect is to take a pretty neutral headphone, significantly improve the bass but to make it otherwise far less neutral, at least to me.

Here’s one example of what I described above. On the HD800S, Sonarworks measured the treble peak between 6-7 kHz, while Oratory measured the same peak nearly an octave lower. Which is correct?

 

[Phorize]

My approach is to use whatever correction most closely matches my headphones to my calibrated monitors. I know this is not a satisfying answer, but I do think it describes the reality of the situation.

I agree it's the only realistic approach short of setting up a proper headphone measurement rig, which is not practical in most situations.

 

[Zensō]

I agree it's the only realistic approach short of setting up a proper headphone measurement rig, which is not practical in most situations.

And even having a measurement rig only mitigates for a portion of the many variables. I feel this is one of those rare instances where measurements have only limited usefulness and using one’s ears is still very much required.