DanielleFrantz-MusicalExpression - Musical Expression No...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Musical Expression No Longer Exclusively Human By: Danielle Frantz Papers A Survey of Computer Systems for Expressive Music Performance Neural Network Based Systems for Computer-Aided Musical Composition:Supervised x Unsupervised Learning YQX Plays Chopin Outline Introduction A survey of Computer Systems for Expressive Music Performance. Computer Aided Musical Composition YQX Plays Piano Ethics Musical Terms Tempo - The speed at which a song is played Measure - A segment of a song separated by bar lines. Dynamics - Represent the volume at which the notes are played. Articulation - Refers to how the notes are played (short, long, smooth, etc.) Vibrato - Fluctuation in pitch in order to produce a more pleasing sound. Introduction to Systems for Musical Expression 1980's - Midi Robotic, Metronomic Sounds Why do they need expression? Instead of making it humanlike, we turned robotic.(Synthpop) Robotic music was RARELY used in classical music. Expressive Performance Actions Refers to the actions of musicians NOT in the score. Tempo and Loudness (NOT in score) Expressive Performance Actions Examples: Vibrato, Intonation, Slowing at the end of a piece, articulation WHY? Embellishment, Internationalism Performance Context Expression of a certain mood or emotion Happiness, Sadness, etc. Change tempo and dynamics to change the emotion. CSEMP Computer Systems for Expressive Music Performance Automated - Once Trained, it generates a new performance Semi-automated - Requires manual input. (Musicological Analysis) Computer Expressive Performance Why would we want a computer to give an expressive performance? Develop computational models (Musicology, Psychology) Composing tool Expressive Computer Generated Music (Generate mood music for a computer game) Playing data files (MIDI) Computer Accompaniment (Track and Generate Expression) Outline Introduction A survey of Computer Systems for Expressive Music Performance. Computer Aided Musical Composition YQX Plays Piano Ethics Framework for Research in CSEMP Modules Performance Knowledge (Rules controlling the system) Music/Analysis (Analysis of structure: Automated, Manual, or a Combination) Performance Context (Emotion or Style) Adaptation Process (Develop performance Knowledge) Performance Examples (Past performances) Instrument model (Post commonly the piano) Methods of CSEMP Nonlearning Linear Regression Artificial Neural Networks Rule/Case-Based Learning Statistical Graphical Models Other Regression Methods Evolutionary Computation Non-Learning Director Musices (Project since 1982) Phrase Arch (Rainbow Shape) Groups Tempo and Dynamics Sound trails off toward the end of a group Works in Real Time Hierarchical Parabola Model Relates Tempo and Dynamics similarly to DM The Louder the Faster Not Convincing Bach Fugue System Conditions Generated from an annotated knowledge base Works for polyphony (polyphony is a texture consisting of two or more independent melodic voices) Commercial Projects Sibelius Program for music composition Built-in algorithms for expressive performance Rules and Techniques Higher the louder Use bar lines and beam groupings to determine emphasis Random fluctuations added (More human-like) Notion and Finale More advanced Fewer details about methodology Learning Methods Can incorporate more knowledge more quickly Not as much creativity Focus on past performances In order for a CSEMP to learn to play expressively, it must be trained with both robotic and expressive performances in order to learn the deviations. Linear Regression Assume linear relationship between music features and expressive actions Advantage: Simplicity Disadvantage: Almost certainly oversimplified Music Interpretation System (MIS) Generates expressive performances in MIDI format Learns expressive rules from audio recordings Links between music features and Performance actions Tempo, Dynamics, Articulation MIS was trained on the first half of a Chopin waltz and generated an expressive performance for the second half. Neural Networks Inspired by the structure and processing of the human brain The Network contains neurons connected together by synapses. Each connection has a weight (Referred to as the interconnection weight) Training Phase Feed the network examples Testing Phase Generate guesses based on the training Set Supervised Vs. Unsupervised Learning Supervised Learning Input-Output Mapping Unsupervised Learning Only Inputs Known Neural Networks Start with random weights. Goal is for the weights to converge to a line separating Red from Green. Outline Introduction A survey of Computer Systems for Expressive Music Performance. Computer Aided Musical Composition YQX Plays Piano Ethics Introduction Neural-Network Based Systems for Computer-Aided Musical Composition Compares Supervised Vs. Unsupervised Learning Uses 'Nature' as inspiration Landscape Contours are converted into integers Focus on Pitch and Duration Nature representation Measure represents 4 time units. Eigth Note: Quarter Note: Half Note: Whole Note: Integer Representation Representation of Pitch Combine integers and intervals Integers For Example: Middle C is 0 Each half step above middle C gets a new positive integer Each half step below middle C gets a new negative integer Intervals Distance between two notes + or - half steps Pitch and Duration Supervised Learning Training Samples: Brazilian Folk Music Trained with measures Once trained, compose new melodies Supervised Using Back-Propogation The network output is compared to the desired output. The difference of the two represents the error, which is then used to adjust the weights. Repeated until some stopping criteria... Example. Threshold or some number of iterations. Unsupervised Learning Unsupervised using a self organizing map (SOM) Cannot calculate error in the training stage. In each iterations, a winner neuron is chosen to update its weights. Weights are updated by adding the Learning rate times the difference between the input and the winner. The neighbors of the winner neuron are also updated by using the Euclidean distance between the winner and its neighbors. Outline Introduction A survey of Computer Systems for Expressive Music Performance. Computer Aided Musical Composition YQX Plays Piano Ethics Overview Contest to create a system to learn to play piano expressively Computer has to play two pieces 'expressively' that it has never seen before The two pieces were composed specifically for the contest (in the style of Chopin) Scores contain normal markings, but no indication of phrasing Contestants have 1 hour to set up their computer, read the scores and generate a performance. They cannot hand-edit the performance or even listen to the generated MIDI Graded by human audience and composer Video Machine Learning Predicts three expressive dimensions timing, dynamics, articulation Now for a Quick Video! Outline Introduction A survey of Computer Systems for Expressive Music Performance. Computer Aided Musical Composition YQX Plays Piano Ethics Ethics and Implications While there are many positive benefits that will undoubtedly help musicians, there is also the possibility of Computers replacing the musicians. If computers can play pieces perfectly and invoke an emotional response, why would we want to listen to a human performer? Computers might be cheaper than human performers. References "A table, a chair, a bowl of fruit and a violin; what else does a man need to be happy?" -Albert Einstein References: Gerhard Widmer, Sebastian Flossmann, and Maarten Grachten. YQX plays Chopin. AI Magazine, 30(3):35– 48, 2009. Kirke, A. and Miranda, E. A Survey of Computer Systems for Expressive Music Performance, ACM Computing Surveys, In Print, 2008. Correa DC, Levada ALM, Saito JH, Mari JF (2008) Neural network based systems for computer-aided musical composition: supervised x unsupervised learning. In: Proc. of ACM Symposium Applied Computing SAC’08 ...
View Full Document

This note was uploaded on 07/30/2011 for the course COP 4810 taught by Professor Staff during the Spring '11 term at University of Central Florida.

Ask a homework question - tutors are online