crossfeed

[mkeroppi]

I love the rockbox crossfeed; it's the only one that sounds right for ER-4B. I want to use it as a plug-in for foobar or winamp, unfortunately, it seems to be in assembly and there's no way I'll be able to decipher it let alone the dsp. Is there more information on it?

I've tried OHL's VST (http://www.ohl.to/audio-tools-and-thoughts/audio-softwares-and-vst-plugins/crossfeed-and-eq-for-headphones/) which is suppose to be the same? The problem is it's not the same even with the same parameters. It's definitely "too bright" like all other crossfeeds. For those who are not familar, the ER-4B is diffused field equalized and is "too bright" except live recordings and of course sounds right for binaural recordings.

The crossfeed is enough for me to use rockbox even though it keeps crashing on the x5. Thanks all.

[xlarge]

It crashes on your x5? What build? It works nice on mine... I'm on cvs:060822-1222

(i'm afraid you have to go somewhere else for info on how to make it work in winamp - not quite the scope for rockbox)

[markun]

There is a C version of the crossfeed routine in dsp.c:
It starts on line 547 and on line 153 you have the C defines to go along with that.

[jlo]

I've tried OHL's VST (http://www.ohl.to/audio-t...ed-and-eq-for-headphones/) which is suppose to be the same? The problem is it's not the same even with the same parameters

I'm a bit surprised because, really, it should be the same when parameters are set the same.
Have you tried latest version ?
The rockbox crossfeed was implemented by preglow with schematics and parameters that I gave him. The default values are the one that I put in my VST software. Haven't you found a setting in this crossfeed VST that suits you ?
Crossfeed isn't supposed to be used for binaural recordings but for normal recordings it works fine. For normal recordings, what kind of EQ do you use for the ER4B ?
This is a very interesting subject and I'd like to work on my crossfeed to find a good solution also for ER4B.

[mkeroppi]

What is Rockbox's "delay" in the crossfeed ?

[jlo]

What is Rockbox's "delay" in the crossfeed ?

Delay is the time delay between the left signal arriving at left ear and same signal arriving at right ear (ipsilateral and the contralateral signals), default value is 14 samples (about 0.35ms), which corresponds to delay between ears when listening to loudspeakers at about 30°. For more information , see :
http://www.ohl.to/audio-tools-and-thoughts/audio-softwares-and-vst-plugins/crossfeed-and-eq-for-headphones/

[mkeroppi]

I mean the actual delay used. I've tried 14 on the VST and it doesn't sound quite the same. Looking at dsp.c doesn't seem to point to anything.

[jlo]

I'm not sure but I think delay is hardcoded in dsp.c

/* Wrap delay line index if bigger than delay line size */
        if (++di > 12)
            di = 0;

So I think delay is 13 samples (not 14, sorry).
Preglow will be able to confirm

[mkeroppi]

That looks right, missed that part.

I'm not getting it why either. The Rockbox crossfeed definitely has a much cleaner sound, vocal sticks out more (the center channel). The algorithm looks simple enough though (maybe the LP filter is better implemented for the purpose?).

Right now using 6 samples delays (for best clean center versus imaging) and -5dB@10kHz EQ (again to clean the sound a bit) on the VST. It seems even unclear center makes things sound "bright". Other settings are the same as the default Rockbox crossfeed setting. Rockbox crossfeed just sounds "right" (imaging) default setting and without EQ, and cleaner.

[mkeroppi]

Had some time and did some tests with the same file VST processed crossfeed off versus crossfeed on the same Rockbox, and can confirm the difference. It's definitely the difference of large clean layered sound of ER-4s but with a roomy mix versus smaller, and less cleaned, squashed sound. The room+details are distinctly good (usually one of the other), and I think that's what drawn me to Rockbox, and it's for doing transcriptions.

I wanted to go ahead and port the Rockbox code, but wanted to see if you can share your code, as I have done DSP in school, but not VST.