HOMEWORK 9B (due Wed, Mar 24)
University of Southern California
ISE 575 / EE 675 / CSCI 575


PREAMBLE

The audio files you need for this homework can be found at www-scf.usc.edu/~ise575/d/audio .  This page is password protected.  You should log in with username: moon, password: beam.

The audio files are:

     Beethoven's Moonlight Sonata

     a)  Daniel Barenboim - moonlight_adagio_barenboim.mp3

     b)  Maurizio Pollini - moonlight_adagio_pollini.mp3

     c)  Artur Schnabel - moonlight_adagio_schnabel.mp3

     Bach's Cello Suite No. 2

     a) Pablo Casals - bach_cellosuite_VCasals.mov

     b) Yo-Yo Ma - bach_cellosuite_YoYoMa.mov

There are also two youtube videos of Gould and Ashkenazy playing the Moonlight Sonata

     a) Glenn Gould - http://www.youtube.com/v/WD6pGV69fJI

     b) Vladimir Ashkenazy - http://www.youtube.com/v/1PqnruPitzc



ASSIGNMENT  [ 150 points ]

You have each been assigned a piece to analyze.  For you analysis, please follow the instructions below.  Much of this will be review of what you have done in previous homeworks.

In all the plots required in [1] and [2] below, mark the mean tempo/loudness, and the one and two standard deviaion intervals around the mean.  Whenever the x-axis is in beats, label the x-axis every three beats.


[1] Tempo

     (a) Use Sonic Visualizer to annotate the beats in your piece.  For both the Beethoven and the Bach pieces, assume that eighth notes form the beat level.  Create a plot of the inter-beat-interval (in seconds) vs. beats.  [ 25 points ]

     (b) Create a plot of T_1, the instantaneous tempo (in beats per minute), vs. beats.  Index each tempo value by the number of the beat that is ahead in time.  [ 20 points ]

     (c) Create a moving average plot of T_3, the tempo (in beats per minute, smoothed over three beats), vs. beats.  Index each moving average tempo value by the number of the beat in the center.  [ 10 points ]

     (d) Create a weighted average plot of the tempo, T'_3(i) = 0.75 * T_1(i) + 0.25 * T_1(i-1), vs. beats.  Label the x-axis every three beats.  [ 10 points ]

     (e) Refer to your answer in part (d).  Predict the onset time of the beat i+1 based on the tempo at the previous beat, T'_3(i).  Determine delta, the amount by which the actual beat i is ahead of its predicted onset.  Plot delta (in milliseconds) vs. beats.  [ 25 points ]


[2] Loudness

     (a) Use the ma_sone function to obtain the loudness values in sones, and create a plot of loudness (in sones) over time (in seconds).  [ 20 points ]

     (b) Use the beat onset information from [1](a) to compute the maximum loudness within each beat.  Create a plot of maximum loudness vs. beats.  [ 25 points ]


[3] Temp-Loudness

     (a) Use your results from [1](c) and [2](b) to produce a time series plot of smoothed tempo vs. maximum loudness at each beat.  [ 15 points ]



E.C. Mar 19, 2010