Post-Seminar

PhD Entry Warning: Can cause drowsiness and extreme boredom. Maybe even projectile vomiting. Proceed at own risk.

Alhamdulillah, my seminar on Friday went well. Buat pertama kalinya dalam sejarah selama aku bergelar student, slides siap ahead of time, with time to rehearse and go over the material, despite keadaan yg x brp mengizinkan, spt yg di entry lalu (woo... ramai family di Msia x puas hati wooo sbb aku tulis entry camtu. Semua backing Aisyah! X aci hehe).

But it wasn't easy. It meant getting up at 5.30 every morning to push myself to work ahead of time, and also coming home at 9 pm on many nights, working over the Xmas holiday and weekends too. I guess when you know you don't have that much time, you tend to use your time more wisely. Wise, of course, is another word for tua lah kan basically... obvious sgt zaman2 muda dulu sungguh lah byk buat aktiviti membuang masa itu hehehe.. shhh!!!

Anyway, I am very happy when I got the feedback drpd my SV dan jg rakan2 se-lab (majoriti nya post-doc and ada yg buat second PhD), that they think I am on the right track, the work so far had been well-constructed, with a rather intensive LR. God knows what a huge relief to hear that, Syukur! My SV added that he's gonna pull some strings and arrange it so that I don't have to do another presentation just for the transfer, bcoz this one seems sufficient already, yeay!


A few points utk membaiki diri di masa hadapan, though:

1 - Don't go too far back with the LR (e.g. earlier work form the 1940s), because there has been so many exhaustive work on them, and you are just putting yourself in real danger that way --> In my defense, I did this bcoz the field is so foreign to me, and I thought I needed to explain all the things and facts I find fascinating from the beginning of time, padahal to other researchers yg dah lama berkecimpung in this field, it might be old news already

2 - Try to limit myself within the scientific part of things, and not to sway too much into musicology, or electro acoustics, because at conferences, when the target audience is large, there is a chance that the musicologists will start to debate on things that are not the main focus of my research, i.e. my definition of music (there are hundreds, and people have actually gotten a PhD just discussing that topic alone!), therefore putting me in a spot where I don't want to be.

3 - Be more confident! Aku ni jenis yg suka nervous2 and sengih2 mcm kerang busuk, pastu kalo org tanya je terus panic2, mata meluas, dada berdebar, mulut kelu dsb nya. One of my labmates said, 'You don't have to be so nervous. People ask you questions because they genuinely want to know the answers from you, not to test you'. Beliau menyambung lg, 'You don't have to agree what whatever people suggest or say' (ye, aku admit aku terlalu cpt to agree, and x defend sgt kerja sendiri), and he went on, 'Remember, this is your work and YOU are the only one who knows the AREA better than anyone else in the room, no matter how many PhDs they have' (eh, betul juga tu!). Advice of the day yg paling kaw2 ialah, 'If you don't know the answer or you're not confident enough about something, FAKE IT'. Wah wah wee wah!

Insya Allah menjadi ingatan dan panduan di kemudian hari. Jelas I have an issue with self-confidence, and dah menjadi kebiasaan kot org Melayu kita ni terlalu modest and selalu underestimate diri sendiri, a negative quality yg harus aku kikis sedikit demi sedikit.

Another exciting news though, is that my SV reckons that it is now time to start my test and simulation with the different audio features. Woo hoo! Finally, something from my MSc years is actually making a comeback. This ultimately means a LOT of work are to follow after this, but we all know that with tests, you get results, and with results, you have some academic contribution to the field, and it means a chance to get your work published as well. My SV is nudging me to start writing a review paper or an introductory short paper to low impact conferences with what I have at the moment, and aim for the bigger ones once the results are here *passed out, but smiling ;P*


Things to do in the nearest future:

1 - Repeat exact same experiment as previously done on the prototype, but with larger dataset (~100 songs)

2 - Test using the features that based on the human perceptual model - spectral centroid, spectral flux, spectral roll off, low energy, zero crossing rate, sum of beat histogram and mean pitch. Compare results one by one. If required, use Principal Component Analysis (PCA) to do this

3 - Test with a slightly different methodology. Assume that the system will perform targeted concatenation, where the sounds in the database are pre-sorted into their respective clusters prior to the selection stage, makanya this part bole la ambil chunks and findings from my work on Musical Genre Classification during my MSc days


Itu la sahaja buat masa ni. Byk tu nak kena buat tu. Plus on Monday my SV will return my report to me, ada minor pembetulan lg kata dia, but by next week Insya Allah dah bole submit to the Faculty. Doakan x de aral apa2 ye. Lama dah rasanya buat report ni.

Oh, saja nak share a clip that I played in my seminar, bcoz the man explains it so much better than I ever could, just so that you have an idea of what this field is about (kalo nak tahu la). This is an actual BBC footage on concatenative music during the beginning of its emergence (in the 70s tau, where the magnetic tape player was then the 'new' technology, sila ketawa terbahak2 okeh). Obviously, everything is done digitally now, but the original idea is still the same. It's about composing the avant-garde music. As expected, the clip discusses more about the arts itself, rather than the science or 'intelligence' behind the work (but that's where I come in nyeh nyeh), but still, a good introductory clip.