What we can learn from the BBC, Pandora, and Musiclens for mass customization
The traditional compilation CD is deader then ever. Instead of listening to all songs of an entire CD, most consumers today prefer just to listen to what they want on their MP3 players (a typical long tail phenomenon). But selecting, filling and arranging the playlists of these players have stayed more or less a craft business. While some persons feel joy and achievement once they have generated their very own, individual playlist, this is, none the less, plenty of work. Also, listeners are restricted to the music they known (and, more or less, own).
Sure, with satellite radio, there are now highly focused radio stations which substitute general radio broadcasts. But often, even these stations are still a bit too broad and contain songs that you don't really like too much – and too few of your real favorites.
Here, three new services provide help. Using different approaches, they allow users to customize their music experience beyond the restrictions of ownership, information about favorite songs, and the demand to manually craft a custom playlist. These services provide tools to find new music matching an individual's preferences, but also enable custom broadcasting services of a new level.
Note: I am not talking here about the option of customizing the particular song. While there are some new promising offers (e.g., at http://www.dabreakupsong.com you can create a custom rap song to break up with your partner :-), this is a minor field of application (see my posting on this subject). Most people don't want to customize their music on the level of the single song, but they want to personalize the stream of songs they listen.
BBC to Develop Personalized Radio Service
Madeforone recently reported about a new personalized radio service that the BBC is developing. The UK state broadcaster wants to allow audiences to create personal radio stations from its content, its director general has said. The service, provisionally called MyBBCRadio, aims to give audiences more control by combining existing services such as podcasts and the BBC Radio Player. It will be part of the BBC’s iPlayer, a new interface device that shall transport custom content (music, video, reportings) to each user.
With its earlier Backstage offering, the BBC has been a forerunner of user-generated content. It changed its policy from protecting its content to giving most of it away to listeners for free, allowing users to create new works by mixing their own stuff with BBC programming (more information here). NNC Backstage, however, was more an offering for leading-edge users or music lovers. But the new MyBBC Radio Service wants to bring this capability to the mainstream.
Radio 2.0: Pandora and the Music Genome Project
But personalized radio stations do already exist. Once of the best services is Pandora, a music discovery service designed to help users find and enjoy music that they like. Based on a huge database that has categorized songs of over 10,000 different artists based on unique attributes, it helps users to find music that has the same characteristics of a song or artist they like. Just type in a name of a favorite song or artist, and let the magic begin. I was highly fascinated by the quality and scope of the resulting personalized music stream (you need an US ZIP code to use this service, if you live abroad, just use 02138, my postal code).
Pandora has a totally different approach to configuration compared to the majority of other configuration toolkits. It is a good example of a need-based system, i.e. an expert system that does not demand that users can describe exactly what they want, but that just analyses what they like and provides suggestions based on this analysis.
The service is powered by the Music Genome Project. In this project, founded by Tim Westergren in January 2000, a group of musicians and music-loving technologists came together with the idea of creating a comprehensive analysis of music. They assembled hundreds of musical attributes or "genes" into a very large "music genome". Taken together these genes capture the unique and musical identity of a song -- everything from melody, harmony and rhythm, to instrumentation, orchestration, arrangement, lyrics, singing, and vocal harmony.
This kind of expert analysis of all songs provides the main difference of Pandora to other music recommendation systems based on collaborative filtering (as, e.g., at Amazon.com). Collaborative filtering works without any idea of the nature of the articles it is recommending, as it is solely based on comparing the usage (shopping) behavior of one user with the behavior of other users. Pandora, on the other hand, is not based on such a analysis, but solely on the nature of the products (songs) it recommends.
Pandora is offered for free in an advertising supported version (and of course the company hopes to get a provision when you buy a song you discovered though its service). There is also a subscription based service without advertising. All music is streamed, this means it is not possible to download or safe the stream (legally). With a special device, you can also listen to your very personal music channel without your computer.
More information and analysis: http://blog.pandora.com/press/
MusicLens: Configure Your Music
Now compare Pandora to MusicLens.de, a German project that aims to test four different technologies to analyze music like it has been done in the Music Genome Project. However, this service opens the Pandora box and allows you to really configure the music you like. Using more than 10 sliders, you can describe exactly the music you want to hear, like its tempo or subjective volume. You can also define the purpose of the music on a scale from listening over driving and sex to dance. And set the mood of the song (on a scale from smile to angry).
As a result, you get a play list with songs matching your desires. Sounds too complicated? You also can just provide a favorite song for the start, see its characteristics, modify them just a bit, and get new music. As this is a test project, however, the music available does not match the scale and scope of Pandora's repertoire. But the web site provides already a good indication of the service.
MusicLens uses a fuzzy search technology to find music - CDs, albums, song titles or artists - by characteristics. Searches can be carried out in large masses of data quickly and effectively. And, despite the terms of the enquiry being vague or indefinite, the results are clear. The idea is to provide users also good results when they do not know what they are looking for – a feature conventional search engines do not have. Users also do not need to use specific search vocabulary or any form of literal definitions. The search is conducted by way of various characteristics or categories, represented by the slider navigation system.
DDD-Systems, a Hamburg based IT services company that is behind MusicLens, hopes to position the system also as a business-to-business service. The search technology shall assist those involved in purchasing film, radio or television content to find the right content.
Beyond segments and clusters
Why do I write about all this? In my opinion, MyBBCRadio, Pandora, and MusicLens provide some great general insights for mass customization:
-- These services overcome the traditional categorizations (market segmentations) of music in genres. There are no clusters of "Independent", "Pop", or "Rock" music. All categorization is based on an individual user's preferences and desires.
-- Pandora and MusicLens apply modern search technologies which support a need-based configuration approach. Instead of today's dominating parameter-based configuration approach (users configure products based on concrete selections of components and modules), they help to define products based on a description of the user's needs and preferences.
-- They address the customization of services. Up to today (and also in this blog), mass customization is often solely discussed in relation to the individualization of physical products. MyBBCRadio, Pandora, and MusicLens are great examples of mass customization of service offerings.
But for now: Enough words, now go ahead and listen to your personal music.
Comments