Sound and Music Computing
A+
AY24/25 S1
Prof is so chill and open to feedback. Definitely one of the most goated teaching teams in SoC.
This mod may be put in the media focus area, but it certainly is HEAVY on deep learning concepts. Automatic Speech Recognition (ASR) and Music Transcription (AMT), audio analysis and synthesis, are some of the core topics covered. You'll also have to think in terms of the frequency domain, as you'll work with fourier transforms (converting signals into the frequency domain), and spectrograms (how frequency components change over the course of the audio clip). You'll learn about different music formats, mainly working with MIDI for the most part.
You'll need a strong understanding of how they work, because they will become the building blocks for your deep learning models. Cos dimensions are always a painful thing to reason with for any model.
There is a heavy emphasis on class participation, as even lecture attendance is expected. And you should. Cos Prof invites external speakers to talk about their areas of expertise. They are pretty good and Prof Wang Ye is very particular about who he invites.
I think the most effort put in by the teaching team was organising the concert near the middle of the semester. For extra participation points, you can perform a piece of music in front of the CS4347 audience. If you can, then you are encouraged to do so. I can't perform directly, so I proposed to arrange one of the pieces. I still got the points anyway. Free food, free concert, goated Prof. Some more it came from his own pocket, so all the thanks to him.
Workload was reasonable for the importance of the concepts taught. I think Prof values thoughtfulness and effort over anything else, so keep this in mind when doing the assignments.
A1 was a critique on one of Prof Wang Ye's past papers. For this, you need to discuss summary, strengths, weaknesses, and your own personal take on it. Try to be unique and you'll get a high score.
A2-4 are labs based on the lecture. A2 is on fourier transforms, A3 on AMT, and A4 on ASR. They are HEAVY, but nothing too out of the ordinary. Half the points are for the code and results, while the other half is for the write-up, which are like short-paragraph-response-questions. The labs are pretty well done, though I feel too much deep learning knowledge is assumed.
A5 is about writing a project proposal and video. For this, make sure you fully know the rubrics, as the proposal is graded based on that and not really on quality. From there, you get to network on the day of the concert to find your groupmates. Do a project relating to S&M Computing, can be research, software, anything. Then write a report, do a video presentation, and you're done. No finals!
Loading comments...