[stv014]

Quote:

Very nice work stv014. Have you considered using valarrays? IME they are faster and have certain arithmetic features that make them more appealing than vectors in certain situations. It's much easier to dynamically grow vectors though.

 
One easy way to improve performance would be to replace my FFT code with the Ooura FFT library, which is also used in SoX, for example. It is about 1.5 times faster. Also, in the resample utility (which is currently about 1.8 times slower than the sox "rate" effect at similar settings), there are some possible changes for additional speedup: the FIR lowpass filter for integer power of two resample ratios could take advantage of the fact that there are zero samples in the input (upsample), or samples of the output will be discarded (downsample). Additionally, the interpolation could use a pre-calculated lookup table for simple rational resample ratios, rather than calculating the sinc function and window directly, as long as the memory requirement of the table is reasonably small.

 

[ultrabike]

That sounds kool. For FFTs I usually go for FFTW libraries. Dunno how they compare to the Ooura library though.

 

[stv014]

FFTW is even faster, but is more difficult to replace the original code with, and it also has a GPL license.

 

[preproman]

Sorry to bump this thread for such an old headphone.  Are there any CSD plots for the JVC DX1000?

 

[ultrabike]

Quote:

Sorry to bump this thread for such an old headphone.  Are there any CSD plots for the JVC DX1000?

Haven't seen any yet, but here is the FR...

 

[vid]

http://www.geocities.jp/ryumatsuba/hp-dx1000.html
 
Though I've heard grumblings about the reliability of the measurements on that site. Check out some of their other measurements to get a feel.

 

[preproman]

Thanks guys
 

 

[vid]

(Yamaha HP-50S, modded.)
 
Worth keeping in mind that in the '70s Yamaha handed these out for free with their line of electronic organs. The phones' main contribution as stated on the box was to enable one to play the organ at night without being a nuisance to others.

 

[Pianist]

I am not sure if this has already been discussed, but it looks like AKG significantly improved on their K550 with the newer K551 - the latter seems to have a more neutral frequency response and a much faster decay. Here are the measurements for K550 and here's K551. The newer model measures so well according to that site, that I am really tempted to buy myself a pair.

 

[vid]

Quote:

I am not sure if this has already been discussed, but it looks like AKG significantly improved on their K550 with the newer K551 - the latter seems to have a more neutral frequency response and a much faster decay. Here are the measurements for K550 and here's K551. The newer model measures so well according to that site, that I am really tempted to buy myself a pair.

 
Looks like they improved on it a bit, and at the same time it looks like it rings in the treble much the same as before. (Though not to say I don't appreciate AKG putting actual effort into an xx1 model rather than just renaming the old one as before.)
 
Mind you, I don't know why Golden Ears use massive 1/3 octave smoothing on their CSD.

 

[jerg]

Quote:

 
Looks like they improved on it a bit, and at the same time it looks like it rings in the treble much the same as before. (Though not to say I don't appreciate AKG putting actual effort into an xx1 model rather than just renaming the old one as before.)
 
Mind you, I don't know why Golden Ears use massive 1/3 octave smoothing on their CSD. Below is how the HP-50 measurement I just did would look like with 1/3 octave smoothing.
 

 
Not very useful, eh.

I think we can all guess why. Reduced smoothing shows a lot more information, likely a bit too much (for them).

 

[purrin]

1/3 octave resolution is about as much we can make out with music. The application of 1/3 octave smoothing is appropriate for frequency response graphs - and for practical application of GED or PEQ.
 
For CSDs where the intent is to locate resonances (seen as ridges), I prefer to use no smoothing or maybe at most 1/9 octave. I feel it's important to see the nature of the ridges without smoothing because there is a difference in how they correspond to what we hear. Also, as you have demonstrated, some of the minor resonances tend to disappear into thin air.

 

[Pianist]

Maybe I don't know how to interpret the graphs properly, but K551 frequency response and csd look better to me than those of HD800 from the same website. Of course, HD800 has a better driver...

 

[vid]

 
 
More Unipolar fun. Raw graph, so the ~3 kHz peak doesn't belong. The others do, unfortunately.
 
The other channel rings a lot less thanks to modding, but this channel seems to resist.

 

[vid]

AKG K 260 as a Philips rebadge. Posted a measurement of this in 2d six months back, though I think it's worth seeing in 3d HD.