real world columns: sound

The underlying model of a tape for a cassette deck is the same for almost all cassette decks; a physical entity carrying two distinct linear sound blocks. True, a cassette tape does have distinct tracks (songs) recorded on it, but the distinct tracks are not known to the system. It knows only the two sides of the tape and the beginning and end. This rather monolithic model is changing as technology improves the underlying model. CDs and CD players know about tracks and can do interesting things with them. But the era of cassette decks was interesting because, although the underlying model of the tape was the same, different decks provided different controls to the user. These controls embodied actions and the key thing was that the user’s goals were achievable with these actions. Consider ‘rewind’; standard decks offered a rewind action. Contrast this with in-car cassette decks where the cassette was inserted sideways with quite a large part of it sticking out. These decks only had the actions ‘play’ and ‘fast-forward’. To achieve the user goal of rewinding a cassette, a composite action was required, namely; ‘turn cassette over’ and ‘fast-forward’. Another less common example is the child’s cassette deck that my daughter has (and that I frequently borrow for lectures!). To achieve the goal of playing at different volumes it doesn’t have a play action and a volume control it has three distinct actions / buttons; ‘play quietly’, ‘play normally’ and ‘play loudly’. Different actions but the same goals are achievable. Sometimes the technical limitations mean that the user is obliged to carry out tricky, composite actions in order to achieve simple goals. Finding the beginning of the next track using the actions ‘play’ and fast-forward’ is possible but very unwieldy. With a video recorder it is even worse, as the video recorder needs to do a lot of whining and clicking between each change of ?operation.

Nowadays the underlying model has been improved with the introduction of sound technologies such as DAT, CD and MiniDisc. These support the idea of separate sound chunks, yielding improved actions (finding the beginning of the next track is a doddle) and a host of new actions like ‘shuffle’ (play all tracks in a random order) and ‘play intros’ (play just the first five seconds of each track). The ‘play intros’ action migrates well to other sound-chunk technologies such as skipping through voice mail messages, while applying a ‘shuffle’ action on your voice mail would definitely not be a useful operation. The most recent, interesting development in this area is the ‘digital wallet’. A box that is, funnily enough, about the size of a cassette tape. It can store several gigabytes of sound tracks (or photos or anything else digital for that matter). Here you can store not just your favorite tracks for the day but your entire music collection; that whole shelf of CDs in one unit. This is interesting because it breaks down the entity of ‘an album’. You can apply a shuffle function to your entire music collection and get all sorts of strange combinations of tracks from different artists and different genres. It’s something like having your own radio station playing only music from your own collection.

The final gem to come out of this overview of music technology is the idea of ‘technology overload’. This is the practice of extracting advanced features from media whose underlying model does not support those features. Dictaphone users spent ages struggling with the problem I alluded to earlier – of finding the beginning of tracks – until one manufacturer introduced a system where the user could press a button to insert a beep as a marker onto the tape, and had actions to skip between these markers during playback. This is a simulation of the separate tracks that are an integral part of the underlying model of formats like CD and DAT. Another example were the cassette players that searched for the start of tracks by playing through quickly to themselves and ‘listening’ for the gaps between tracks. Again imitating the separate track formats of CD and DAT, but this time not extending the format with user-inserted markers but applying low-level AI to the existing format. However, like all low-level AI it was not 100% perfect, and the cassette system I had as a student played havoc with my Joni Mitchell tapes, turning some of her solo voice tracks into multiple tracks with just one verse in each!