[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [EXTERNAL] Re: phantom sources

It is well known that there is substantial sensitivity to ITD for signals at all frequencies. At higher frequencies, the information is encoded in the envelope rather than in the carrier. This follows directly from the inability of the system to encode the phase of the carrier at higher frequencies – an issue the envelope encoding avoids.


Some references:

Henning, G. B.  1974   Detectability of  interaural delay in high-frequency complex waveforms. J . Acoust. Soc. Am. 55, 84-90.

Bernstein, L. R., Trahiotis, C., 2002. Enhancing sensitivity to interaural delays at high frequencies by using “transposed stimuli”. J. Acoust. Soc. Am. 112, 1026–1036.





Frederick (Erick) Gallun, PhD
Research Investigator, National Center for Rehabilitative Auditory Research

Portland VA Medical Center

Associate Professor, Dept. of Otolaryngology and Neuroscience Graduate Program

Oregon Health & Science University


From: AUDITORY - Research in Auditory Perception [mailto:AUDITORY@xxxxxxxxxxxxxxx] On Behalf Of James Johnston
Sent: Sunday, July 27, 2014 10:51 PM
To: AUDITORY@xxxxxxxxxxxxxxx
Subject: [EXTERNAL] Re: phantom sources


Well, having read more than one paper saying that there are no time cues above 2kHz (or 3kHz, take your choice) I offer everyone the following challenge.

Take a pulse train at maybe 100Hz.  Filter it with a standard elliptic HP so that there is no energy to speak of below 2kHz (or 3, your choice).

LIsten to it time synced in stereo.

Listen to it with a 2 sample delay in one channel, identical signals, identical levels, i.e. copy with 2 sample delay into one channel.

In speakers, tell me that they don't sound different.
In headphones, ditto.

I have recently seen some scathing, and completely wrong, reviews arising from some 1955 tests that attempted time-delay pan, using increments of 5 milliseconds.

Rather obviously, unless your head is 5 feet wide, 5 milliseconds is excessively too much.

So, be careful with the literature out there, rather some of it can be disproven with a very simple matlab example.



On Sun, Jul 27, 2014 at 3:21 AM, Enda Bates <enda.bates@xxxxxxxxx> wrote:

Hi Philip,

         yeah there's been quite a bit of work on this type of thing in audio research. You are also basically correct in that a phantom source will be a little broader than a real source, largely due to the fact that the localization of phantom sources is somewhat frequency dependent. Straightforward amplitude panning will produce a slight discrepancy between the localization at low and high frequency ranges, which leads to a slight change in timbre (manifests as a slight spectral dip at approx. 2kHz) and an increase in the Apparent Source Width (ASW) of the source when it is reproduced as a phantom image. 


This is a big problem when dynamically panning a sound as this change in timbre tends to highlight the loudspeakers in a negative way and it significantly reduces the smoothness of the spatial trajectory (incidentally, one the main goals of the Ambisonics spatialization system was to eliminate this issue, which it largely does but not without some costs). 


Ville Pullki has done a lot of research in this area, particularly in terms of his amplitude panning system VBAP. He has also conducted research on panning using three loudspeakers (again with VBAP). My recollection is that this will increase the ASW even further, and similarly reduce localization accuracy. Of course there is a strong relationship between the two factors, and the concept of locatedness is useful in this regard. 


Anyway, chapter 6 my PhD thesis has lots of specific references in terms of listening tests with phantom sources and the relationship between localization accuracy and apparent source width, both for stereophony and ambisonics. You can find it here -> http://endabates.net/academic.html


Hope that helps,

Dr. Enda Bates

Music & Media Technology,

Trinity College Dublin








On 26 July 2014 16:45, William Yost <William.Yost@xxxxxxx> wrote:

Fine work in Journal of the Acoustical Society of America by Wes Grantham

Sent from my IPhone

William Yost

Office: 480-727-7148



On Jul 25, 2014, at 9:17 PM, "Philip Robinson" <philrob22@xxxxxxxxx> wrote:

Howdy y’all,


I’m doing some investigations preliminary to implementing a room acoustics auralization system and want to get some input from the community.


Is there any good literature on the perceptual difference between a phantom source (e.g. between a stereo pair) and a real source (e.g. center channel)?  I am interested in more than just simple localization: width, resolution of multiple phantoms, etc.? Is perception of a phantom source the same when panned between a triad or a pair of loudspeakers? My basic understanding is that phantom sources are wider or blurrier than real ones, has this been specifically tested?


Any opinions or references are welcome.


Have a great summer,


Philip W. Robinson 

Environmental Design Analyst

Specialist Modelling Group


Foster + Partners

Riverside, 22 Hester Road

London SW11 4AN

E  probinson@xxxxxxxxxxxxxxxxxxxxx





James D. (jj) Johnston

Independent Audio and Electroacoustics Consultant