[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Auditory wheel
Hello Michael and the list,
Thanks Malcolm for pointing out our work!
I recently submitted my PhD thesis which discusses the perception of
the sound color (i.e. spectral energy) and the density (i.e.
fine-scale temporal variation of sound pressure) with their
quantitative representations. Either or both of these qualities can
constitute a seamlessly morphing timbre space.
I wish I have had my thesis online but Stanford is still preparing
the electronic dissertation database
I will let the list know once it is officially uploaded. For now, I
can send you my thesis file so please let me know if you're
** Hiroko moved to TARA Center, University of Tsukuba!
On Fri, Mar 19, 2010 at 10:48 AM, Malcolm Slaney <malcolm@xxxxxxxx> wrote:
> On Mar 18, 2010, at 5:22 PM, Michael H. Coen wrote:
>> I'm looking for the auditory equivalent of a color wheel. Namely, a
>> parametrized, continuous method for generating a series of sounds that
>> form a "perceptual loop" that has no perceived gaps.
> Hiroko Terasawa did some work on this problem a few years ago.
> The color wheel works because it shows all colors in a perceptually relevant
> space. It's just a three-dimensional (or four if you are blessed) space and
> you can easily move between points in straight lines. It's a complete
> There is no equivalent in the auditory space, yet.
> The first two dimensions of an auditory space appear to be pitch and
> loudness, and then something related to timbre. Terasawa showed that
> low-dimensional cepstral coefficients are a good representation for static
> timbre sounds. With such a representation one could smoothly move from one
> sound to another in timbre space. The papers are online at
> Unlike previous work on timbre spaces, she started with a sound synthesis
> procedure and then measured how parsimonious the sounds were with respect to
> a linear perceptual space. That space might be close to your desired wheel.
> Let me know if you have questions.
> - Malcolm