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

Re: [AUDITORY] Tolerance time interval for piano performance recognition

Hi Federico,

I wrote my PhD thesis about "tolerance time interval” (a.k.a. simultaneity window, or I call "perceptual simultaneity range”) in 2018.
I measured the time interval not for piano sounds but for two pure tones. However, below may be helpful to you.


S. Okazaki and M. Ichikawa, “Perceptual simultaneity range as a function of frequency separation for two pure tones,” Acoust. Sci. & Tech., 38(4), 185–192 (2017). https://doi.org/10.1250/ast.38.185 Fig. 5
S. Okazaki and M. Ichikawa, “Effects of frequency separation and fundamental frequency on perception of simultaneity of the tones,” Proc. Meet. Acoust., 29, 050004 (2016). https://doi.org/10.1121/2.0000474 Fig. 3


In short, the tolerance time interval will be changed by frequency separation and F1 frequency.


Cheers,
Satoshi Okazaki

OKAZAKI Satoshi, PhD
 |  JSPS Postdoctoral Fellow
 |  Dept. Music, Kyoto City University of Arts
 |  Email: sat.okazaki@xxxxxxxxxx



> 2019/03/28 20:37、Federico Simonetta <federico.simonetta@xxxxxxxx>のメール:
> 
> Dear list,
> I am a Ph.D. student in music informatics at the University of Milan. My project is about piano performance analysis and score-informed piano transcription.
> 
> I was wondering if someone here knows any study about one or more of the following time tolerance intervals. I am interested in the threshold after which a human can recognize that in piano performances:  
> 
> * two or more onsets are not synchronous
> * two onsets are in different timing positions in respect to the previous identical note offset/onset
> * two or more offsets are not synchronous
> * two notes have different durations in monophonic/polyphonic environments
> 
> I have found the following related paper, but it is rather old:
> E. F. Clarke, “The Perception of Expressive Timing in Music,” vol. 51, no. 1, pp. 2–9, Jun. 1989.
> 
> From this study, it seems that humans are able to recognize differences in music performances even if time changes lasts only 20 ms. However, most of the researches involving computational analysis of music performances (audio-to-score alignment and automatic music transcription), refer to the threshold of 50 ms as tolerance.
> 
> I am wondering if, as of today, some more recent and in-depth research has been carried on.
> 
> Thank you very much to anyone's help!
> 
> Cheers,
> federico
> 
> ---
> 
> Federico Simonetta, PhD student
> 
> LIM - Music Informatics Laboratory 
> Dept. of Computer Science 
> University of Milano 
> Via Celoria 18 
> I-20133 Milano - ITALY 
> 
> Skype: federico_simonetta
> http://www.lim.di.unimi.it
> http://federicosimonetta.frama.io