Help - Search - Members - Calendar
Full Version: Dolby Headphone
Hydrogenaudio Forums > Hydrogenaudio Forum > Uploads
SebastianG
Hi, Folks !

This is a stereo impulse response of the "Dolby Headphone - Live" setting which I generated via:
- encoding a unit pulse (left channel) to AC3
- playback of AC3 via PowerDVD with DolbyHeadphone enabled
- recording the output

Unfortunately the Foobar2K convolver plugin does not support cross-feeding and I don't know any convolver plugin for any player that supports this kind of processing that is needed to simulate DolbyHeadphone.

This is how you have generate the correct output
CODE

given: s, the audio signal to be processed (complex: real=left, imaginary=right)
given: r, the impulse response (complex: real=left, imaginary=right)
out: p, the processed audio signal (complex: real=left, imaginary=right)
---
p = convolve(real(s),r) + swap(convolve(imag(s),r))
---
real(.)        returns the real part of the complex signal
imag(.)        returns the imagunary part of the complex signal as real signal
convolve(.,.)  convolves 2 complex signals and returns the result
swap(.)        swaps the real and imaginary part, swap(x) = i*conjugate(x)



Sebastian


Dolby Headphone for Foobar Thread

edit: corrected definition of swap(.)
kode54
I would recommend redoing that with DPCM, but then you're probably limited to 48000Hz.

Also, it might be a good idea to attempt the 5.1 result for a unit pulse on each of the input channels, but then you'll need a cross convolver.
SebastianG
QUOTE(kode54 @ Sep 11 2004, 01:57 AM)
I would recommend redoing that with DPCM, but then you're probably limited to 48000Hz.

Also, it might be a good idea to attempt the 5.1 result for a unit pulse on each of the input channels, but then you'll need a cross convolver.
*



I think the quality is ok for impulse responses. The (quite low) noise that got introduced by the recodring process is as white as the pulses which leads to a little extra reverb when convolved with the signal - but this is very likely to be impercible.

Here's another "impulse response set"... the Zip (66KB) contains
- 5pulses.wav (5 channel, 48 kHz WAV, 16 bits)
- 5pulses-640kbps.ac3 (the same in AC3 format)
- dolby-headphone-ref-48k-LEFT.flac
- dolby-headphone-ref-48k-CENTER.flac
- dolby-headphone-ref-48k-RIGHT.flac
- dolby-headphone-ref-48k-SURROUND-LEFT.flac
- dolby-headphone-ref-48k-SURROUND-RIGHT.flac

This time i used the setting called "Reference" which I believe simulates a medium room with a 5.1 speaker system.

The .1 channel does not need an extra impulse response. The center-IR can be used for the LFE channel as well. (This is what the Dolby Headphone software does)

Have fun,
Sebastian
mkeroppi
This is cool, and seem to work wonders on the er-4b on normal recordings (I no longer get headaches seemingly searching for the sound location subconsciously). There seems to be one problem. The center channel sounds chorusy, is that possible because the left and right are mixed to form the center, and these channels are actually a little out of sync? It doesn't seem to have this effect when using just the left or right channel.
bug80
Why don't you use a sweep signal as input and deconvolve afterwards? That way, you'd get a much better SNR. You have a servere amount of noise on your measurements.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.