Yes, hard to see the burst decay in time scale. convert it to periods, then it’s easier to read. Or use the Impulse response.
Yes Group Delay is certainly good too.
Yes, I’m a quarter way watching the video yesterday. I’ll watch it. Thank you
@Resolve, on this graph, the 104.1dB is at 14.84kHz?
I’m curious about if and how time sequences may affect timbre. For example, an acoustic bass may generate a similar FR as a given drum, but they sound very different. Also, a wooden clarinet generates medium-high tones similar to a tenor saxophone.
Regarding music reproduction, dynamic speaker drivers have timbres (to me and others) that parallel the materials they are made from. Paper cones sound papery, plastic cones sound plasticky, and metal cones sound metallic – plastic vs. metal is super obvious when comparing the Sennheiser 6 series to Focal’s range. Whether or not this is measurable (or how it may be measurable) I do not know.
1 Like
That would all be in FR at the ear drum. So we can’t say that the FR at the DRP would be the same without actually measuring them in-situ at the DRP, and this is also one of the major sources of differing impressions on a headphone. We can’t say the indicated response on the graph is how it’s going to measure at the DRP for any individual because headphones behave differently on different heads. They can be close, but not exactly the same, and they can also be quite different depending on the headphone’s design.
Regarding the timbre question, that’s also FR, in particular the Focal’s often have a midrange bump and then some fine-grained treble features. But yeah the idea that paper cone drivers = papery or metal driver = metal sound, that’s difficult to really assess without establishing what constitutes ‘papery’ or ‘metal sound’. Once that’s established then we can figure that out in the FR analysis too.
It would if you were looking at an impulse, and yes technically if you sweep high enough, it’ should all be covered in the FR.
Obviously most headphones don’t have to deal with phase issues because of driver alignment or a crossover. And there is a 1:1 correspondence from time domain to frequency, but that doesn’t mean a simple fixed volume sweep (often at volumes that aren’t normal listening) is enough to capture all driver behavior.
The problem I think is the assumption that you can capture FR with a frequency in isolation and that captures how it behaves with a multitude of inputs at different amplitudes.
Or perhaps in the frequency domain it’s just hard to see the relevant information.
What are you suggesting would change here? You do get different behavior for harmonic products but if they’re below the audible threshold then it doesn’t really matter. Are you suggesting the actual frequency response would change? This doesn’t ever seem to be the case.
Edit: You can also get the same result with an FFT.
Tap on a cardboard box = papery (dampened highs, minimal sustain)
Tap on a plastic box = plasticky (thick, boomy, low-middle emphasis)
Tap on a metal box = metallic (sharp, bright, substantial sustain)
Record these sounds, and have it with your FR and other measurements
Compare box measurements to headphone driver measurements
If the driver were perfectly stiff, and they aren’t, and the driver infinitely fast, it wouldn’t ne an issue.
This is one of the reasons paper cones on 2Ch speakers sound very different than plastic cones, even though in most cases the crossovers roll off their contribution before they start to struggle.
The mass of the cone also contributes and matters even with a crossover, limiting it’s frequency contribution.
You potentially get effects like large amplitudes at one frequency partially masking another.
If a driver can’t instantaneously represent a single frequency, but say takes 2 or 3 cycles, you won’t catch that in a sweep, because it’s steady state response could be fine, and your incrementally changing what your asking the driver to do.
Sure I was using FFT’s over pink and brown noise to measure and address room responses 25 years ago.
I understand the correlation, both time and frequency domain information are equivalent for a given test, but the reasons that you look at somethings in the time domain aren’t always because it’s not represented in the frequency domain, but rather because some things are more easily interpreted in the time domain.
1 Like
Yeah and being able to see certain things given that alternative view of the data may be of some benefit in headphones, although I haven’t personally found any of that more useful than just seeing the effect in FR. But I think the big distinction in general has to be drawn between headphones and speakers for some of these metrics, and far too frequently they get used to indicate things they don’t actually explain.
And, I’m not sure this is exactly the point you were getting at earlier but at a recent CanJam, @Mad_Economist gave a presentation on how you can get the same FR as you do with a sweep from music. This was to address the common complaint people have about measurements that “but music is more complex than test tones”. We’ll need to publish that video at some point so folks can reference it, but we may as well use the sweep or an FFT and just look at the frequency view. And importantly consider that heard features that don’t show up on the graph are still bound to be FR-related (when things aren’t totally wonky), just a difference of ‘heads’.
1 Like