Re: [linux-audio-user] artificial-head IR-file for jack_convolve

From: Andrew Gaydenko <a@email-addr-hidden>
Date: Sat Dec 10 2005 - 22:19:26 EET

Bungee,

I don't say about stereo sound image. The aim is to move virtual sound image
_beyond_ a head. A decision must work with mono stream too (as you are listening
to a single real speaker).

Take headphones and apply any kind of LCR filters - you will get a sound image
between your ears.

Andrew

======= On Friday 09 December 2005 23:23, Bungee wrote: =======
Wolfgang Woehl <tito@email-addr-hidden> wrote:

> Andrew, current edition of the german computer magazine "c't" has an
> article on virtual spatialization. They mention "Dolby headphone (tm)"
> and another effort by Creative (makers of the soundblaster stuff). What
> they say about results perfectly dampens any higher expectations.
>
> Reading the thing makes me doubt that simply convolving with a room
> response would do what you want. "Kunstkopf", a very old technique,
> might be worth googling for. And in the back of my mind there's a faint
> note about David R. Clark (davidrclark@email-addr-hidden) knowing things
> about headphone spatialization. He used to be on LAU.
>
> Wolfgang

I don't know if this helps in anyway but I remeber reading some articles in the
late 70s for 'real world' headphone drive circuits.

One issue was that phones don't allow the normal leakage to both ears at middle
and low frequencies like loudspeakers do. The suggested solution to this was
resistors in series with both channels and an inductor from one to the other so
that at low frequenceis the channels were effectively joined, becoming more
separated at higher frequencies. There was also a lot of discussion surrounding
this about phase differences etc. based on the different path lengths being
about 8 inches. There was some very heavy stuff!

For what was termed as expanding the stereo picture (which is what I think th OP
is talking about) it was suggested that transformers be used to inject a small
ANTIPHASE percentage of the R into the L channel and vice-versa.

I vaugely remember someone producing a drawing (wireless world?) that combined
these features with variable controls. The one that made me laugh was 'head
density'

Presumably, with the appropriate filters these ideas should be possible in
software.

-- 
Bungee
Received on Sun Dec 11 00:15:07 2005

This archive was generated by hypermail 2.1.8 : Sun Dec 11 2005 - 00:15:07 EET