For today you should:
1) Read Chapters 1 and 2 if you have not.
2) Work on exercises.
3) Explore one of the readings or project ideas below.
4) Prepare a progress report (see Lecture 1).
Today:
1) Chapter 1 and 2 review
2) Four experiments
For next time:
1) Read Chapter 3 on NB and make comments/ask questions.
2) Work on exercises.
3) Review case study ideas in Lecture 02.
4) Recommended "reading": http://auditoryneuroscience.com/topics/lecture-cortical-representations-complex-sounds
For today I prepared four experiments:
1) To practice the tools and concepts we have learned so far,
2) To demonstrate the kind of exercises you can create for yourselves, which might turn into case studies, and
3) As a starting place for some in-class explorations.
They are in lecture03.ipynb, which you should get by pulling from upstream.
Let's look at the harmonic structure of a triangle wave and synthesize one.
Explorations:
1) What happens if we get the phase information "wrong". Do we get other waveforms that sound the same?
2) Can we do the same thing with a square wave?
Saxophone is an interesting sound in part because the fundamental tone is often not the dominant tone.
What happens as we delete harmonics above and below the dominant tone?
Explorations:
1) Even after filtering out the harmonics, the result doesn't sound (to me) like a sinusoid. What's still there?
The most obvious way to shift the pitch of a wave is to shift the spectrum. But that doesn't work. Let's see if we can figure out why.
Explorations:
1) What's attempt #2?
The notes of a major chord share many of their harmonics, which raises the question of why and when we hear as sound as a chord, as opposed to a single pitch with harmonics.
Explorations:
1) Can we figure out what's going on at 1:21 of this video?
And around 3:00 in this one:
Also:
Shepard tones
And here are some links from my occasional correspondent Sam Keene at Cooper Union:
This years Signal Processing Cup, looks like fun, and I'll be advising a team from Cooper:
http://www.zhilinzhang.com/spcup2015/
Last years Cup, we attempted this one, but did not fare very well at all, I wasn't crazy about the project, thought it was a bit too hard, but it is image processing:
http://www.icassp2014.org/SP_cup.html
I used this IEEE RWEP project as a intro to DSP and MATLAB last year, the kids liked it, its not too intense:
http://www.realworldengineering.org/index.php?page=project&n=1&project=23
There is a conference, Machine Learning for Signal Processing that lies at the boundary of DSP and Machine learning, so that might interest you. I participated in the latest contest with some students, it was pretty hard:
http://mlsp2014.conwiz.dk/competition.htm#.U-vRhqMjCRk
The previous year contest was also fun, some students did it for a machine learning course project:
http://mlsp2013.conwiz.dk/competition.htm#.U-vRvqMjCRk
I had some students do this contest for a speech/audio DSP contest. We did pretty well on the scene classification stuff. This is another one at the border of signal processing and machine learning. Signal processing for the features, then machine learning for the classification:
http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/
Here is another signal processing/machine learning one. This one is neat because it comes with a full, open source speech rec tool, and the goal is to do signal processing to improve its performance:
http://spandh.dcs.shef.ac.uk/chime_challenge/
This is another one I'm aware of, but haven't done:
For DSP course related stuff, I use the oppenheim and schafer book, which comes with a lot of pretty good MATLAB projects. I think you need to register to get access:
http://wps.prenhall.com/wps/media/access/Pearson_Default/8099/8293581/login.html
A faculty member at Columbia that I did my MS with, has a lot of good stuff online. He puts all his stuff online, and has some good labs:
http://www.ee.columbia.edu/~dpwe/e4896/practicals.html
I should warn you, since you are entering the DSP world, MATLAB is pretty entrenched over Python.