There’s a big science festival this week at Imperial College:
the Imperial Festival. There’ll be “hands-on demonstrations, music, comedy, dancing and art”, and: “Food and drinks are available throughout, just drop in at any time.” I’ve been involved in research on the DarwinTunes evolutionary music generator with some Bob MacCallum and Armand Leroi from Imperial College, and they’re presenting a live version on the Saturday. I’m definitely going on Saturday, so would be great to see you all there.

I’m excited to welcome Dr Klaus Frieler to Queen Mary — he’s the first researcher on a project initiated by me. Together with Simon Dixon, Klaus and I are going to look into how singers stay in tune, and when they lose it. We are bursting with ideas, and so I think the three months Klaus will work with us will allow us to get a better idea on what it actually is that we want to learn about intonation. Usually based in Hamburg, Germany, Klaus has already done quite a lot of interesting projects around singing, for example a funky demo and presentation at the London’s Dana Centre at the Science Museum that showed people how hard it is not to plagiarise. Looking forward to exciting research. Welcome Klaus!

A few months ago I disabled comments on my site — it was just so annoying to get all that spam. Now I’ve started them again, with a simple captcha, but my guess is that I’m going to get fed up soon again. What do you think?

The other week I got the amazing invitation to “represent” the Royal Academy of Engineering at a weird little event called Voice of the Future. In a nutshell, it was a bit like prime minister’s question time, only that we didn’t ask the prime minister, but David Willetts, the Minister for Universities and Science, and other politicians. I put the word “represent” in inverted commas because—in my case—representing meant simply sitting at the table and looking interested. (My fellow Research Fellow Andrew Robertson actually did get to ask a question.) The Speaker opened the event, and though the event didn’t take place in the famous House of Commons, it was certainly quite interestin…

A big thanks to Dan Stowell for making the brilliant Yanno web application, which uses my Chordino Vamp plugin to extract the chords of any YouTube video you like. For example, say you find a video of Help!, then you can click a special bookmark (as explained on the Yanno page) and what you get is a web site with the chord transcription showing alongside the video. Now, everyone who follows the music information retrieval publication circuit just a little bit has ample opportunity to witness that chord transcription is not nearly perfect yet, and Chordino isn’t even state-of-the-art! (It was always going to be only a simple chord estimator on top of NNLS Chroma.) So don’t expect miracles, but it might just be a little bit use…

While lazily browsing wired.com last night I was delighted to happen upon research of my former colleague Kurihara at AIST: the SpeechJammer. While the immediate application is somewhat contrived (and a case of “only in Japan”, maybe), it’s a fabulous piece of research with public impact, and a lot of good comedy elements. Highly recommended.

I have had a great time in Utrecht talking (at a symposium see my slides) and listening to stuff about harmony — quite inspiring at times. Yesterday it was a pleasure to see Bas de Haas (check out his mafioso photo) get his PhD in a public defense ceremony that was quite different to my own (which was two people asking me about my thesis for about two hours in a small meeting room at Queen Mary): audience, jury and the candidate convened in a grand hall clad with paintings of (I assume) the old greats of Utrecht scholarship. The opponents, clad in robes and wearing hats, had the duty to probe the candidate’s knowledge — but in a kind of non-physical show wrestling style, i.e. they were joking, and Bas’s responses seemed (I couldn’t actually understand them) to aim at providing a good show to the audience rather than making a compelling argument. That’s because (unlike in the UK) the serious review and revision happens months before the defense. In any case, very interesting, and well done Bas! Today was the day of the actual symposium, and luckily that was all in English. Much easier, though the talks on grammars and syntax of harmony by Bas, José Pedro Magalhães and Martin Rohrmeier that started the day off were still a bit hard to deeply understand — will have to read up on it in Bas’s thesis.
Meinard Müller gave some nice insights into chord labelling performance as a factor of different kinds of preprocessing and processing steps, whitening, smoothing over time, and different inference techniques, especially HMM vs simple chord pattern. (The main theme of Meinard’s talk, which was the cross-version chord display, which I already knew plenty about.) I also quite liked Aline Honing’s talk on extracting low-level interval characteristics from MIDI, which (after multi-dimensional scaling — sorry for the scary word!) appeared to cluster pieces from different classical periods together nicely, though it remains to be seen how that qualitatively differs from existing algorithms in Humdrum and other symbolic processing toolboxes. Xavier Serra talked about his exciting multi-cultural project CompMusic, and it really made sense to me today: there is really a lot to be learned by stepping out of our own culture, to find that the methods we thought worked do not generally work. I expect lots of colourful new MIR to come our way from Barcelona soon! Similarly Anja Volk presented her new Musiva project on similarity and variation — I’m looking forward to the slides, because she had some really nice citations from the cognitive science literature and beyond.…

I’ll give a short talk on my work on harmony in the past few years next Wednesday in Utrecht. The talk is part of the Symposium on Harmony and Variation in Music Information Retrieval organised by Bas de Haas as a satellite event to his PhD viva. Quite excitingly, some big names of MIR will be there. Thanks to my collaborations I have a lot of juicy stuff to talk about: the SongPrompter application I did at AIST with Hiromasa Fujihara and Masataka Goto, the amazing Songle.jp web service, a big project at AIST to which I contributed chord, key and beat/bar estimates and Last.fm’s Driver’s Seat, a collaboration with Mark Levy and Sven Over already span quite a wide range of MIR applications, and I will also show a bit of Yanno…

After way to much time, I’ve finally managed to upload a OSX universal binary of the NNLS Chroma Vamp library. If you’re using OSX and NNLS Chroma, this should make it significantly faster… thanks to the tech miracle that is compiler optimisation. General information on NNLS Chroma and Chordino can still be found on the Isophonics website, and the open source repository is on the code.soundsoftware.ac.uk website.

It was my first ever Music Hack Day, and despite being sceptical at the beginning, I was eventually won over by the average quality of all the hacks. My positive impression was—let’s say—enhanced by the fact that the hack Sven Over and I made actually won a super duper Spotify prize. Read all about it on the Last.fm blog or check out the video of the hack demo right here.