Sound Signatures and Graphs

V-shaped refers to the left and right parts of the frequency response graph (i.e. bass and treble) being high while the middle parts are lower.

note that on HeadRoom graphs, treble should be lower than flat to sound neutral; imagine the treble higher than it actually shows above

Bright means the treble is stronger, generally. Dark means that the bass (usually mids too) are stronger. The more complicated it gets, the less consensus there seems to be.

There's often more to a sound signature than can be described on a response graph, though it's a pretty good start. For example, a response graph is time-blind, but hearing isn't. Resonances may show up as a small response peak, but are more audible than that because of how long they last (a bit overly simplified, but close). But a 3D "waterfall" graph is actually somewhat hard to digest.  We fall back to descriptive terms, but language lacks many terms that describe sound, so most of the terms what we use are analogous to visual, physical or tactile ideas, and some others convey an emotional sense.  

 

Have a look at this gossary, you'll see what I mean.  What's really hard to do is assign a specific frequency response to these terms.  The terms probably reference a whole family of response curves. And, as you'll see from this thread, the list is incomplete.

 

http://www.head-fi.org/t/220770/describing-sound-a-glossary

Frequency response is the most prominent aspect of sound that differentiates headphones. All other aspects are miniscule in comparison.

I would say that CSD plots are also very important, if not equally so. For example, the CSD plot of a white-box Audio-Technica ATH-M50 reveals that there is mild ringing in the midrange at around 4Khz, which explains why it sounds more neutral than the frequency response graph would suggest. The midrange looks recessed on the frequency graph but the extra energy at 4Khz somewhat makes up for that in real-world listening.

 

The frequency response graph should show the total amount of energy at a particular frequency over the impulse response, it is not just the first frame of the CSD.

I actually wasn't aware that someone was doing that much in CSD (waterfall) for headphones.  When I googled it, it took me right back here!  

http://www.head-fi.org/t/566929/headphone-csd-waterfall-plots

 

If you read the introductory post, you'll see that the goal of CSD headphone testing is to try to better visualize what is heard, an improvement over simple time-blind FR.  Then skip around a bit, and read through some of the results and comments.  What you'll find backs up my earlier statement that waterfall plots are difficult to digest.  The do reveal ringing quite well, but what people hear doesn't always correlate as well as you'd hope for.  A CSD isn't hard to understand.  It's the connection between what's seen and what's heard that isn't as solid as we'd like.  When the CSD supports the audible impression, comments are like "that explains it." "I heard that xKHz ring."  When they don't support the impression, people seem to say, "That's weird" "didn't expect that". It's an interesting mix.  What's also not published is the detail of how the tests were run, unless I've just not looked in the right thread.  

 

Not trying to shoot down CSDs here, they are important, especially when combined with standard FR.  It's just that presenting the raw CSD isn't 100% reliable in determining the signature, more like 70% or less.  I'd tend to assume that rings are more audible when longer/stronger/wider and mid-band, but I'm not sure there's any published tests that reveal audible thresholds, like what was done for the masking curve research for compression codecs.  Someone needs to do a paper on the audibility of ringing components, including magnitude and time duration, bandwidth of the ring, frequency vs audibility, integration with adjacent rings, etc.  You'd probably need a fairly ring-less transducer, and simulate ringing digitally so you'd have control and repeatability across test subjects.

 

OK....go!

I googled "list of sound signatures" and found this.  Without graphs and lesser technical terms, is there a more "ordinary Joe" way of understanding sound signature of the product?  I was thinking that some people mention signatures by the brand.  Although some specific models sway from such signature, it's mostly a little (just paraphrasing what I read from some threads).

 

I was hoping there's a bit more general formula of sound signatures per brand, for those who don't have the opportunity to test each model before buying.  At least a little insight of what to expect.  Something like:

 

Beyerdynamic:
******** bass

**** mids

***** highs

******* soundstage

***** cold

... and whatever else is mentioned characteristics when talking about sound-signature.  Add the "lower mids" and what not as well.

 

Do you guys think there's a way to make a general formula like this?  The way I see it, when people talk about the "v" shape, I just imagine the equalizer forming the shape of "v."  Same goes with "u" shaped.  But looking at those graphs, I don't even see the figures.

 

Maybe I'm just stupid.  Sorry for rambling.

No, you're not stupid for rambling. I have the same question as you when I read "u" and "v" shaped sq. Those are vague terms thrown around here. We all know that we hear differently. And most everyone here is a reviewer to a certain extent. But we can't measure every iem or headphone out there. And there lies the dilemma.

Sonic signatures are like written ones, very individual, and very detailed.  You can talk about them in generalized terms, but those descriptions won't get you the real picture.  In a written signature you can talk about upward or downward slants, loops, curves, etc., but that won't help anyone visualize the actual signature.  

 

People who talk about sonic signatures in terms of V or U response curves are doing exactly the same thing, either from over simplification or lack of understanding.  The response curve is the real key.  To really get the idea of what a signature sounds like from a measured response curve you have to make the connection between indicated frequencies on the graph and what they sound like, then understand the interaction of the width of a peak or dip with it's magnitude.  Basically that relationship is smaller width peaks and dips are less audible than wide ones, and so the narrower the excursion the deeper or higher it can be before it becomes audibly objectionable.  How audible a response variance is affected by where it is in the spectrum also. At the extremes, variations are harder to hear.  In the middle, fairly easy.  

 

The hard part is trying to imagine what a particular curve sounds like.  But for that, we have new tools on our computers, like editors with equalizers that give the user not only a graph of what it's doing but also the ability to modify the width (Q) and degree of boost/cut.  This can help you to audition the sound of those types of response changes, and little by little make sense of the measured curves indicate something about the signature.

 

Of course, frequency response isn't the whole story, but it's the biggest part of it.  If you can understand that, you'll get most of what the signature sounds like.