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

Re: Roughness in audio and vision



Hi Bryan;

I just want to add some additional information to Question 1:

1) Do we have a reliable method to measure the roughness of a given a natural sound or image?

One can calculate roughness or aperiodicities in the acoustic signal by looking at not only amplitude perturbation (shimmer), but frequency pertubation (jitter) as well as noise to harmonics ratio (NHR), degree of tremor, voice breaks (for isolated vowe prolongations), number or degree of sub-harmonics, etc.. Some commercially pre-pared acoustic software programs (by KayPentax, such as their MDVP program allow one to measure at one time a host of 30 or so parameters on one acoustic signal at a single time).

With respect to reliability, published literature suggests that some measurements are more reliable than others for various speaking tasks. Single vowel prolongations, e.g., [a] prolongation, tend to yield more consistency than connected speech. However the idea is, the less jitter, shimmer, NHR, and sub-harmonics, the less "roughness" in the voice and chaos in the spectrogram. Of course, one needs to control the collection of the acoustic signal by recording in a sounded treated room or a room with an ambient noise level of < 50 dB, use a condenser microphone, etc.. I believe the specs to control artifacts in the recording and collection of the acoustic signal can be found on the NCVS website: http://www.ncvs.org/ There are also some great tutorials can be viewed on this website on such topics at: http://www.ncvs.org/ncvs/tutorials/voiceprod/tutorial/index.html

In terms of reliability, there are some issues on the reliability of some of these parameters. There are numerous published articles on this topic. My research team found relatively strong reliability in measuring degrees of roughness in the following articles:

Multimodal Standardization of Voice Among Four Multicultural Populations*1Fundamental Frequency and Spectral Characteristics
Journal of Voice, Volume 15, Issue 2, Pages 194-219
M.Andrianopoulos, K.Darrow, J.Chen

Multimodal Standardization of Voice Among Four Multicultural Populations Formant Structures
Journal of Voice, Volume 15, Issue 1, Pages 61-77
M.Andrianopoulos, K.Darrow, J.Chen

Good luck:

Mary A
UMass-Amherst

----- Original Message ----- From: "Dan Stowell" <dan.stowell@xxxxxxxxxxxxxxx>
To: <AUDITORY@xxxxxxxxxxxxxxx>
Sent: Monday, February 16, 2009 6:01 AM
Subject: Re: Roughness in audio and vision


Hi Bryan -

Bryan Pardo wrote:
Some colleagues of mine are interested in the relationship between roughness in visual images and audio images. They sent me the following questions they were thinking about in hopes that I might be able to provide some references to get them started. I figured this is just the mailing list to get some pointers to papers. If any of these questions make you think of a paper or two, I’d appreciate your emailing the reference.


1) Do we have a reliable method to measure the roughness of a given a natural sound or image?

For images I wouldn't know (maybe some measure of fractal dimension?
http://dx.doi.org/10.1103/PhysRevA.39.1500 ) but for sound, I keep
finding papers where auditory roughness is said to be related to fast
amplitude modulation (AM), e.g.

Joder et al (2009), TASLP
http://dx.doi.org/10.1109/TASL.2008.2007613

which references a thesis I haven't read (Eronen 2001) as the source of
their method for measuring AM in the 10--40 Hz range.


2) How could one synthesis sound clips (and images) with ascending or descending order of roughness?

If that kind of AM does indeed cause auditory roughness then
synthesising is easy, just change the depth of the AM.


3) How can acoustic roughness influence the perceived roughness of the vision?

Good question!

Dan

--
Dan Stowell
Centre for Digital Music
School of Electronic Engineering and Computer Science
Queen Mary, University of London
Mile End Road, London E1 4NS
http://www.elec.qmul.ac.uk/department/staff/research/dans.htm
http://www.mcld.co.uk/