I've been planning to do a little project. It could be called
a user interface experiment. It is somewhat derived from the
thoughts in my previous post about attention.
The basic problem I am trying to solve is the sad state that my
interest in music has been for many years now. I find that I am
not interested in hearing new music, or music I haven't listened
to before. I think this is very much a problem of attention.
The way we are now "supposed" to consume music is "immediately".
This lacks the whole process of finding a record, and maybe not
understanding it at first, until you finally realize how good it
is. Nowadays there is so much music competing for attention that
the market forces would make the record be rather simple, and
more immediately understood.
So, how to compete against this tendency? I know that there is
something magical in the late 60s and 70s music that I could be
interested in, sort of similar way as some people see classical
music. The whole genre around psychedelic rock is fascinating.
But I have not listened to it that much, and would not know much
about the era, musically.
My assumption is that by cobbling together a device with enough
friction, I can bring back the original way of listening to
music, without having to buy an LP player. I can even make it
better than the original way.
Rule 1: The exchangeable media has to keep a year of music on it
Rule 2: The artist / album hierarchy should be broken
Rule 3: The album as a playlist convention should be brought back
Rule one is about simulating the experience of being in certain
era of music. If all the music of that year is on the same disk,
then I will be "grounded" in time. While I may not be able
to remember the names of the songs, I will be able to remember
what sort of feeling that year had. It is also interesting to
see how the music styles of popular artists are being born in
relation to the times they are living in.
Rule two is in a similar vein. If you listen to an artist, one
album after the next, you are speeding up the progress to a
crazy level, and you will most likely end up focusing on their
"best years". What can be discovered by listening to their first
steps without having the immediate access to their successes?
Rule three is about not regarding music as individual songs, but
as a collection of them.
I had plenty of ideas how I might get this user inteface built.
In the end it came down to what pieces of equipment I had and
was willing to buy. I was thinking of the cool version of this,
an mp3 player with exchangeable SD cards, but I didn't want to
spend that much money. I found an old Raspberry Pi B and a USB
numpad laying around and decided to use normal USB sticks
as the media. The numpad can probably be made to take in letters
sort of like an old dumbphone, where you have to press number 2
for three times to get a C.
I first tried a USB mouse as an input, but it turns out that
if you are running a Raspi headless, the mouse buttons are not
useful since you don't have a screen to receive the button
presses. At least that's what I think went wrong. The numpad
is better anyway, since the mouse had only four usable buttons
and a scroll wheel.
As for the output of menus and other information, I am planning
to use Mimic3 to create a "voice assistant". This is actually
a big part of my inspiration, as Mimic3 sounds to me very much
like HAL9000 with it's basic voice.
Running Mimic3 on the old Raspi is not going to work fast enough,
so I am pre-recording all the sounds for the UI. Since I am
confining my record collection to less than a hundred albums per
year, there is no reason why I couldn't record all the
information quite comfortably. I am also recording several
statements for one piece of information to make Mimic sound more
conversational. So, there is a folder of refusals, for example,
where I recorded things like "I'm sorry I can't do that", "It
seems impossible" and so on.
I don't know how the conversational nature of the UI will work in
practice. Will it be too annoying? I think my intuitions about
this might be off the mark, it being a "speaking machine", and so
it is very different from having just an "efficient menu". In
some sense it might need a conversational UI.
