Chris, the other half of the BBC Radio & Music R&D team has built an intriguing new widget to connect up BBC Radio with last.fm. You can read more over at his blog.
It is a Yahoo! widget that lets you listen to the BBC's live radio streams (Radio 1, Radio 2, 1Xtra and 6Music) and then post what is being played to your own last.fm listening profile. If you really like that radio station then you can just let it automatically post everything that's played while you're listening. Or if you'd prefer some control you can use the "Manual" mode and just hit the plus button to manually post what's now playing.
The other cool feature is that it uses last.fm's recommendation technology to show other recommended artists based on the one currently playing - useful if you like that artist but also potentially useful for context if you've never heard of the artist that's playing.
It also looks really nice, subtly including a band photo, your last.fm picture and the radio station logo. Well, not that last one 'cos we're not allowed to include that, but it would if we could. There's a nice picture of a radio instead.
You can download the widget here, but be warned that it is experimental and we cannot guarantee that it will work reliably (or at all!).
Thursday, October 19, 2006
Monday, October 16, 2006
I recently found this paper from Tristan Jehan at MIT on "Creating Music by Listening". It's pretty long but it is so interesting and so ties in with some things I've been thinking about recently that rather than posting a link I thought I'd write up what I understand from it here. Up front I'd like to say that my interest in the paper has nothing to do with the author being my namesake nor the fact that he obviously has good musical taste - he gives examples of his technology applied to Herbie Hancock, James Brown and batucadas amongst others. Speaking of which there are loads of audio examples of the technology on his website and I recommend you listen to these at the appropriate point in the text.
The author of the paper wants to know if computers can be creative and compose music. Although there are principles and rules in the process of composition today these are very complex and it is probably not practical to go down the route of trying to collate, learn and use these. Instead the approach used is to listen to musical examples, attempt to create an internal model of them and to use this to synthesize music. The thesis concentrates on "composing new music automatically by recycling a preexisting one". More specifically it concerns itself with working with polyphonic digital samples of the music on micro and macro scales. One motivation behind the work is to personalise music by potentially providing the listener with exactly the music they want and not being restricted by what has gone before.
The architecture proposed in the paper is basically machine analysis -> machine learning -> machine synthesis. The analysis consists of a number of stages: a listening model representing what is heard by humans, the extraction of musical features, an analysis of the time component of the music (particularly repeating patterns) and an analysis of the macro-structure of the music. The fundamental timescale of these processes is of the order of 10ms, 100ms, 1s and several seconds or more respectively. Most music can be seen to have this structure from beats to patterns to riffs to verses, choruses and movements and structural hierarchies are also seen in frequency as well as time - patterns of notes, chord sequences etc.
If we examine the initial analysis in more detail it looks at features of the music such as loudness, timbre, segmentation into small units (e.g. notes, chords, drum sounds), beats, tempo, pitch and harmony (i.e. chords). Various methods are described in the paper for attempting to determine these from the source audio. For determining the structure of the music the system uses an approach of looking for self-similarities - segmenting the music by time, looking at a particular section and comparing it to all the other sections. The comparison uses the features extracted previously; pitch, rhythm, timbre etc. and works on a series of hierarchical levels of time to find short-term and longer-term structures.
Once the analysis has been done on a piece of music then there are a number of possible applications that the paper looks at including compression, restoration and composition.
Music compression algorithms, like MP3, typically compress in the frequency domain - they remove frequencies that are not audible or are less important. By analysing the structure of music this system could allow compression in the time domain as well. Interestingly, too much compression of this type results in musical distortion while the sound quality stays the same. Obviously this would be most effective on repetitive music, as much modern urban and dance music is.
One application of this technology would be to restore corrupted digital audio - if a section of the music is missing then it can be replaced with a new section synthesised from the rest of the music. This can be made easier and more efficient by including the structural information as metadata embedded in the audio. Again it works best with repetitive music.
The composition aspect of this work is based on what has been learnt from existing songs. The simplest application looked at is an automated DJ doing beat matching and cross-fading between tracks, the examples given are pretty impressive, managing transitions between various tempos and various styles of music. Next up are music textures - the system can take a clip of music, analyse it and then extend it to infinity while never seeming to repeat - the author calls these "music textures". Some of these sounded a bit like the mutant offspring of a skipping CD player while some of them were so good that it felt a bit like being permanently stuck in the intro of a song! Finally the paper looks at music cross-synthesis which creates some kind of musical mash-ups. It takes the musical structure of one piece and the sound content of another piece and synthesises a completely new piece. This is as far as the system goes at the moment but the author's next aspiration is to automatically create entirely new compositions.
So if that was interesting the dissertation itself, some presentations and loads of audio examples can be found online.
The author of the paper is also one of the founders of the Echo Nest. Type a music-related word into the box and it starts playing snippets of sound that seem to be related. I've no idea what this is but based on this paper it could be pretty interesting. (Guardian interview).
Other miscellaneous facts from the paper:
The "tatum" is "the lowest regular perceived pulse train that a listener intuitively infers from the timing of perceived musical events; a time quantum.". Named after the jazz pianist Art Tatum noted for his virtuosity at the keyboard and his long runs of notes.
The James Brown case. Because JB's music is often characterised by a repeated single chord and syncopated rhythms it can be difficult to extract the down beat (i.e. the first beat of the bar) which is typically detected by analysing the rhythm and the chord changes.
Posted by tristan at 15:29
Wednesday, October 11, 2006
Radio 1 launched their redesigned site last night. What's most exciting for me, apart from its excellent new design and implementation, is the inclusion of the SMS cloud (in the panel at the bottom of the page). As listeners send in text messages from their mobile phones the cloud shows the incoming words waxing and waning throughout the day. It's particularly exciting because it's the first time that my R&D team at R&Mi have taken something all the way from idea to prototype to live.
I've written a bit before about my initial thinking and prototyping of this but here's a quick recap:
Many BBC radio networks have numbers that listeners can text in to using SMS. DJs and producers have a "text console" web application which they can use to view the incoming messages and either respond to them on air or select messages to publish to the website. Radio 1 in particular get a large number of incoming messages, between 500 and 1000 messages an hour on average during the daytime, and they've used this in the past for features like the 10-hour takeover. But what I wanted to do was to somehow use all of the incoming messages without needing any human moderation. And we can't just publish all the messages without moderation because there is offensive, private or otherwise dodgy information in some of them. The solution was to try to extract the collective wisdom from all the incoming messages. We do this by filtering the incoming messages to remove any common or offensive words, identify the most popular words at any one time and visualise the top words in a tag cloud. I built a live internal prototype using Processing, you can see a timelapse video of it in my previous post, and showed it around the department.
Since then, the always forward looking Radio 1 interactive team picked up on it, Chris and Lee have implemented it properly (Chris writes about his Flash client here)and it's now on the live site. It's basically the same as my original prototype but a lot nicer and with click-through links on a set of Radio 1-related keywords. Have a look.
It's really interesting how you can get the general idea of what is being talked about on the radio just by looking at these few words; from the weather to the Hoff's new single to football to shopping and so on. I'm slightly worried about what the listeners will do with it once they see they can affect it and I'm expecting some creative combinations of words and some tweaking of the code. But it will also be fascinating to see if and how the DJs interact with the audience using the SMS cloud as a medium. The words also have potential to generate useful metadata describing programmes - particularly around subjects that are talked about, songs that were liked and maybe even moods. I'm sure you can imagine ways to navigate around programmes using these facets. But most of all I like the live feeling of it, how it represents all the listeners out there and how these things are now reflected better on the website.
We're also hoping to get it onto a big plasma screen in the reception of Yalding House (Radio 1's home) and, when we have time, we'll experiment with some other visualisations like timelines and maybe do some more processing of the data. I also intend to provide the raw XML of the popular words for other developers to play with (via BBC Backstage), I'm just trying to overcome the technical hurdles to providing this in some vaguely live manner. Also next is to think about what we could do with incoming picture messages - though I suspect that's several order of magnitudes harder.
Finally, there should be a few more interesting releases from R&Mi R&D before the year is over so keep coming back...
Posted by tristan at 10:35
Sunday, October 01, 2006
Friday was the 60th anniversary of the Third Programme, or BBC Radio 3 as it is now called. There's a BBC site marking this, including a Radio Times from 1946. Last night's Between the Ears on Radio 3 (still available on the Radio Player for a few days) was "Three and a Third", a fantastic compilation of clips from the 60 years of the Third Programme and Radio 3. It was built around the time when one of the announcers, John Holstrom, had to cover 30 minutes of silence as a piano was hauled onto the stage for a performance of a Bartok piano concerto of the Royal Albert Hall for a Prom concert. The programme was also interspersed with many of the pauses, ums, errs and ahs that make up Radio 3.
Particarly interesting was the section of the programme on a new agreement that increased the amount of music that could be played (possibly in the 1960s? The Beatles were playing in the background, the programme used the device of popular music extracts to indicate the time period. Have the Spice Girls been broadcast on Radio 3 before?). It went from a limit of 28 hours to allowing 75 hours a week of broadcasting gramophone records. This puts current discussions on DRM, music rights and the Radio Player into perspective. Worrying about playing more than 28 hours of music per week sounds pretty ridiculous now so how will today's DRM and 7 day window restrictions on music downloads and streaming be viewed in another 60 years time?
"...why should there be any limitation at all?"
"...constant outpouring of music there will be some who are perhaps torn betwen delight at being able to hear fine works at almost any moment of the day and on the otherhand a fear that if music is to become a perpetual companion, if we are to live in an auditory decor as it were, then somehow our delight in it may grow less and music itself will be cheapened...I share this fear just a little myself"
Finally, the last quote from this programme was a very "BBC" announcer deeply apologising for having to remind listeners to pay their television license. Because if they don't there may be
"...ostracism in the supermarket"
"...the deepest shame when walking the dog".
Front Row on Friday also focussed on a long-forgotten aspect of the Third Programme. Every evening before programmes began at five minutes to six it would broadcast the note A (at 440 Hz, the international standard of musical pitch) for 4 minutes as a public service at the request of the British Standards Institute. At the time Dr Alexander of the engineering division said...
"Musicians, musical instrument makers and amateurs now have available in their own homes or workshops an easily accessible standard of pitch and no longer have to depend on tuning forks of often doubtful accuracy."
A quick search reveals that the International Standards Association conference in May 1939 held at Broadcasting House established the frequency of A in the treble clef ('concert pitch') at 440Hz, having previously varied between 435Hz and 442Hz.
Posted by tristan at 12:42