The Influence of perceptual Attack times in
Networked Music performance
Pilot Study conducted at CCRMA, Stanford University (2011)
44th AES conference (2011 - San Diego)
Vector Databases 101 - An introduction to the world of Vector Databases
44th AES conference (2011)
1. THE INFLUENCE OF PERCEPTUAL ATTACK
TIMES IN
NETWORKED MUSIC PERFORMANCE
Pilot Study conducted at CCRMA, Stanford University (2011)
Álvaro Barbosa, João Cordeiro
Research Center for Science and Technology of the Arts
CITAR (UCP–Porto, Portugal)
2. Background on Network Music Performance
Latency and Networked Music
In a vocal conversation it is possible to maintain it even with one-way
delays of up to 500 ms.
Holub, J., Kastner, M., Tomiska, O. 2007
In order to maintain a synchronized and smooth musical interaction
reduces drastically to the order of tens of milliseconds.
Schuett, N. 2002
Chafe C., Gurevich M, 2004
Lago, N and Kon, F. 2004
Barbosa, A., Carôt, A. 2005
Chew, E., Sawchuk, A., Tanoue, C., and Zimmermann, R. 2005
Bartlette, C., Headlam, D., Bocko, M., Velikic, G. 2006
Farner, S., Solvang, A., Sæbo, A., Svensson, U. P. 2009
Chafe, C., Cáceres, J-P., Gurevich, M., 2010
3. Background on Network Music Performance
Latency and Networked Music
For the Human ear to perceive the order of two simultaneous sounds,
they should not be displaced in time over 20ms (Hirsh, 1959)
A difficulty in discerning the order of sounds events
hard to maintain a synchronous musical interaction.
4. Background on Network Music Performance
Latency and Networked Music
The ability to perform music synchronously is strongly dependent on:
• The music expressive qualities (Dynamics and Articulation)
• The music style (rhythm, melody, harmony)
• The music perceptual qualities (pitch, texture, timbre)
• The music structure/form
• The musician’s experience and practicing strategies
• Complementary feed-back modalities (visual, tactile)
• The listening conditions
5. Background on Network Music Performance
Latency and Networked Music
Basic Principles
(1) Ensemble Performance Threshold (EPT)
(2) Echo Feed-Back (Self Delay)
(3) Inverse Proportion to Tempo
(4) Reverb and Complementary Modalities
(5) Slow Attack Times
6. Background on Network Music Performance
(1) Ensemble Performance Threshold (EPT)
(CCRMA 2004, USC 2005)
7. Background on Network Music Performance
(2) Echo Feed-Back (USC 2005)
Chew, E., Sawchuk, A., Tanoue, C., and Zimmermann, R. 2005
9. Background on Network Music Performance
(3) Inverse Proportion to Tempo (UCP-Porto 2005)
10. Background on Network Music Performance
(3) Inverse Proportion to Tempo (CCRMA/Banff 2006)
St. Lawrence
String Quartet
(over 50ms Delay)
11. Background on Network Music Performance
(4) Reverb and Complementary Modalities
Studies on the effect of Reverberation
(Farner 2009)
The Influence of Delay and Various Acoustic Environments
Studies on the effect of Visual Feed-Back
Pilot experiments at CCRMA since 2006
Studies on the effect of Haptic Feedback ???
12. The Influence of Attack Times in NMP
(5) Slow Attack Times
Bregman (1990)
Auditory Scene Analysis
This experiment clearly reveals that
it is harder to perceive the order of
overlapping sound events when
their rise time (attack) is slower.
In other words, overlapping sounds
with slower rise times are better
perceived as synchronous even
when their onsets are not
physically simultaneous.
13. The Influence of Attack Times in NMP
Perceptual Attack Time (PAT)
The fact that two sounds are physically synchronous does not necessarily
make them sound perceptually synchronous.
Perceptual Attack Time (PAT) is a subjective measure of the time that is
perceived by the listener as the moment of rhythmic placement for a
musical sound event.
According to Matt Wright
“for highly percussive sounds, the PAT might be the same as, or just a few
milliseconds after, the onset time, but for sounds with a slow attack, for
example, bowed violin, the PAT might be dozens of milliseconds after the
onset”.
14. The Influence of Attack Times in NMP
Pilot Study Hypothesis
If slow attack times allow a better perception of sound simultaneity this
might also lead to a better ability to perform synchronous musical
interaction
Experimental Setup
15. The Influence of Attack Times in NMP
Rhythmic Performance
80 BPM (to avoid possible overlaying of subsequent notes)
Slow Tempo => Higher EPT
Range from 25 ms to 110 ms (with steps of 10 ms)
28 trials for Slow and Fast attack strokes
16. The Influence of Attack Times in NMP
Results #1
Could each musician maintain a 80 BPM steady rhythm, throughout the
range of latencies introduced on the feedback loop?
We performed the analysis of the average BPM progression with the
“Tempo and Beat Tracker” tool from Queen Mary’s Sonic Visualizer.
17. The Influence of Attack Times in NMP
Results #2
How each musician performs in relation to the other?
analysis was performed using the software tool MATCH: Music Alignment
Tool Chest, developed by Simon Dixon (this software analyses the
alignment of audio files using the OLTW algorithm)
18. The Influence of Attack Times in NMP
Conclusions & Future Work
1. The results from the “Attack Time Pilot Experiment” suggest that for
higher latencies than the generally established EPT (20-40 ms), a
better musical performance can be achieved when musical notes are
played with slow attack times.
2. Further investigation remains to be done, specifically by expanding the
experiment to a larger number of subjects and a larger scope of
musical instruments.
3. It remains to be determined what is if there is a functional relationship
between the rise times of Attack in musical notes and the performance
Tempo, that affects the EPT.