Saturday, February 26, 2011

Aural Timetables: A new tool for connectivity analysis?

A key topic here is transport timetables, especially where there are opportunities for legibility, frequent service or service co-ordination.

Those needing to make a connection typically have to juggle two printed timetables, or, more recently, use an internet journey planner which does a similar thing after comparing alternatives.

In both cases arrival and departure times are presented in visual format.

The visual approach is good for planning specific trips. It could be numerical (eg calculating connections from two columns of times, often in different booklets) or graphical (as used in train graphs, and I think proposed by Tufte).

However I'm unaware of anyone converting timetables to sounds and analysing the result by listening.

Yet aural timetables could offer some fresh insights. For example it provides a way of comparing multiple timetables simultaneously for patterns; something the eye is poor at as it can only look at one time at a time and make a subtraction.

The possibility of aural timetables has evolved from earlier analogies made between some audio/musical concepts (eg resonance, harmonics, time, beat and pulse) and transport timetabling (at least for timed transfer systems). The term 'harmonised headway' hints at this relationship.

More recently I built a drone machine as used by electronic musicians. This is a bank of variable tone generators that can make many interesting sounds and beat notes. I then saw the possibility of audio frequencies being determined by bus and train times at a particular interchange, unwittingly turning service planners into composers.

Such an instrument would be more advanced than my simple drone machine, most likely having a master oscillator and dividers to accurately provide for specific service frequencies. You would also need to shift the start and finish times for each route to reflect actual timetables.

Another approach is to do without special electronics and simply use an audio editing program. I used Audacity which features an audio tone generator just right for this purpose.

The video below explains the concept.

Unlike future videos this one is based on a hypothetical location to introduce the concept of a multi-route aural timetable likely to be useful for connectivity analysis. It starts off with a train arriving every 15 minutes. To this is added a feeder bus every 30 minutes, then another every 60 minutes. At this point the local network is harmonised and the rhythm is distinctive. Then a non-harmonised 50 minute frequency route is introduced and the rhythm (and regularity of connections) is upset. The per second pulse is restored when this service is rescheduled to run every hour. For passengers this means predictable waiting times and an effective frequency for trips involving connections no worse than that of the least frequent route.

You may need to replay this several times. First listen to tones of the same frequency to discern their interval. Once these are identified, listen to the intervals between tones (easiest at lower frequencies, eg 50 or 60 minutes).

Most significantly, note is how the 50 and 60 minute headway services start as being far apart, then arrive at once in the middle, then space further apart near the end. These actually follow a long cycle based on the lowest common multiple (ie a cycle length of 300 seconds here or 300 minutes in the real world).

Such cycles are too long to discern aurally, but a further speeding up may be required to pick this up. Conversely slowing down might help with higher frequency services.

The speed of playback affects how the mind sees these noises; either as individual beeps or rhythms different for each combination of service frequencies. This is similar to Morse code. At slow speeds (eg 5wpm) one must count dots and dashes whereas at higher speeds (eg 20wpm) each letter has a distinctive sound. The advantage of this is increased by the mind's limited memory (or persistence) of sound and memorised rhythms require less effort to distinguish and decode than counting dots. It is painful for a person proficient in the faster speed to return to the lower speed method of decoding.

The same applies to timetables converted into sound. Particularly marked is that the harmonised timetable example (15/30/60/60 minute frequencies) is easier to listen to than the non-harmonised example (15/30/50/60 minute frequencies). I think this is because the harmonised pattern has a cycle recurring every 60 minutes (1 second) whereas the non-harmonised pattern's cycle only repeats every 300 minutes (5 seconds). A series that repeats every second is easy for the brain to sync to, whereas a 5 second long rhythm would be much harder.

What are the strengths and weaknesses of listening to (rather than looking at) timetables?

For precise connections between particular services I still think that either one of the visual approaches are better. However when it comes to examining headway compatibility at stations or interchanges, the aural approach has advantages. Here we are examining the state of connectivity by literally taking its pulse.

Tuesday, February 15, 2011

New Metro train timetables

Starting on May 8.

Major changes for Werribee, Altona, Williamstown, Frankston, Alamein, Glen Waverley and Blackburn trains, with more running direct to Flinders Street and a consistent off-peak cross-city loop between Frankston, Flinders St, Southern Cross and Werribee/Williamtown.

Altona and Williamstown shuttles swapped on weekdays.

Overhauled peak period timetables, with more services on many lines (making use of the new trains).

Off-peak frequency changes: Werribee down from 2 x 20 min to 1 x 20 min, but Newport up to 1 x 10 min.

Major weekday evening service frequency upgrades for Werribee, Altona, Williamstown, Pakenham and Cranbourne trains (30 to 20 minutes).

More at Metro Trains and Herald Sun