Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Interactive Networked Music --Tapping into the Internet as anAcoustical / Musical Medium
Chris Chafe
Center for Computer Research inMusic and Acoustics (CCRMA)
Stanford University
Our theme as a magazine cover...
(ACMDec-01)
(Dec-06)
Stanford Chicago
RPI, NYSanta Cruz
Analogies to other acoustical media: air, water, earth, etc.
topologies
Acoustical Monitoring for Network Operations(NASA)
Interactive Sound Art / Theater(SF MOMA, Parc de la Villette)
Distributed Musical Ensembles
Distributed Instrument Synthesis (Physical Models)
Distributed DSP Algorithms (Internet Reverb)
Applications:
Studies:
Studies of Rhythm Locking under Delay Conditions
Thoughts on the Qualities of an IP Acoustical Medium
Duo PianoConcert
Stanford / Belfast
(15-May-07)
Duo PianoConcert
Stanford / Belfast
(15-May-07)
Jam SessionStanford ?? (2002)
next studio?Berkeley?
Sacramento?Denver
Chicago
Montreal
London
Bass & Saxat Stanford
Listening quiz: The drums are where?
Jam SessionStanford ?? (2002)
next studio?Berkeley?
Sacramento?Denver
Chicago
Montreal
London
Bass & Saxat Stanford
Listening quiz: The drums are where?
Drums are inMontreal
Bass & Saxat Stanford
Jam SessionStanford McGill Univ. (2002)
Jam SessionsStanford McGill Univ. (2002)
Round trip time (ms)Stanford Princeton
Quintet (June, 2006)Banff, Alberta (St. Lawrence String Qt.)CCRMA, Stanford (2nd viola)
(25 msec oneway)
Round trip time (ms)Stanford Princeton
Listening from Banff side
Round trip time (ms)Stanford Princeton
Listening from Stanford side
Quintet (June, 2006)Banff, Alberta (St. Lawrence String Qt.)CCRMA, Stanford (2nd viola)
Round trip time (ms)Stanford Princeton
The Perfect Ritard
Round trip time (ms)Stanford Princeton
The Perfect Ritard
Egads! A new use forthe Internet.
Round trip time (ms)Stanford Princeton
Fighting off the Ritard
Round trip time (ms)Stanford Princeton
Fighting off the Ritard
Hmm... wishful thinking.
Round trip time (ms)Stanford Princeton
Power Structure, the 4 ignore the 1(music minus one)
4 players
1 player follows
Round trip time (ms)Stanford Princeton
Power Structure, the 4 ignore the 1(music minus one)
4 players
1 player follows
Best! ...but result is onesided.
Here's what happensjust across a room... (length of room, ca. 20 feet)
4 players
Quartet, recorded in a room.Take 1: all together in one corner
4 players
Take 2, separated by a partition
3 players
exiled violist
Take 2, separated by a partition
3 players
exiled violist
Take 2, after recording,compensated by 21 msec
Take 2, after recording,compensated by 21 msec
Summary:
Separation delay is like molasses (slows tempo down)
Equality of ensemble makes it worse
Tactics to compensate (“y ou lead, I'll follow”)
Post-production compensation
Room and reverb seem to help
Let's make some music, anyway.
Trio (May, 2004)Stanford (bass guitar) / Victoria, BC (sitar) / Missoula, MT (elec. Violin)
Round trip time (ms)Stanford Princeton
Quartet (June, 2004)Stanford (flute, elec. cello)Stockholm (bass flute, piano)
...really slow tempo,not much rhythm...
Round trip time (ms)Stanford Princeton
Quartet (June, 2004)Stanford (flute, elec. cello)Stockholm (bass flute, piano)
...really slow tempo,not much rhythm...
Round trip time (ms)Stanford Princeton
Octet (March, 2006)Stanford (violin)Seattle (uilleann pipes)Belfast (sextet)
...fast, rhythmic...
Round trip time (ms)Stanford Princeton
Octet (March, 2006)Stanford (violin)Seattle (uilleann pipes)Belfast (sextet)
...fast, rhythmic...
Round trip time (ms)Stanford PrincetonNext gig?
Experiment Designed toDetermine Effect of Latency onEnsemble Accuracy
Subjects = students and staff at Stanford(paired randomly)
Task = play rhythm accurately, keep an even tempo(no strategies given)
Interlocking rhythm
Sound(2ms delay each direction, metronome cue = mm94)
Delays: 0 – 77ms (each way) in 12 steps
Experiment 1 with 17 pairs of subjects
0 25 9
1420
2735
44
54
65
77
Delay Times (ms) Tested
Deceleration from longer delaybut where does it start to cause
trouble?
Sound(77ms delay each direction, metronome cue = mm90)
Deceleration from longer delaybut where does it start to cause
trouble?
Sound(77ms delay each direction, metronome cue = mm90)
Results
Human clappers at 65ms delay
Acceleration vs. delay time(r2 = 0.98)
Tempo slope = 0
Delay = 11.5 ms
@ Delay 11.5 ms, 85% of trials slowed down
Sweet Spot?
Rhythm Locking
new field of study, useful for applicationssimilar to turn taking studies
we like to lock, music, dance, ...physical, fun
clumsy locking is frustrating
and tech stuff can really mess it up(i.e., bad network audio) but, also human stuff
Analogies to other acoustical media: air, water, earth, etc.
speed of sound
due to network delay jitter and asymmetrythe speed of sound is never uniform
(but there are also special LAN digital audio networks using dedicated hardware which are constant)
never uniform
easy to interpret by ear...
sonified pings Stanford www.newscientist.com
never uniform
easy to interpret by ear...
sonified pings Stanford www.newscientist.com
sound quality
depends on QoS of network pathand hosts' network and audiocapabilities
...test with audio feedback loops
fast echoes = pitches
SoundWIRE “ plucking” technique, recirculates audio for fine-grained QoS monitoring
Stanford UCBerkeley
changing delay = variable speed of sound = potential signal discontinuities = sonic booms
packet transmission errors = glitches from dropped data or duplicates
anisotropic
in wood, speed of sound isfaster with the grain
in network, speed of sound is faster in the corebut slows down near the edges where there's often lots of routers
Round trip time (ms)Stanford Princeton
asymmetric
speed never exactly the same bidirectionally
Best match: < 25ms = longest delay > 25 ms = mean of delays
Asymmetry
no distance effects on signal = no attenuation, filtering or dispersion
no mixing =wave channels can be 100% discrete
SummaryInternet acoustics, same but differentMore than just a telephone because there's computing involved...musical instruments, spaces, effects are created in software...and prediction algorithms
Thanks!