[LAD] Experience driven design and Linux Audio

Len Ovens len at ovenwerks.net
Thu Oct 2 21:22:33 UTC 2014

On Thu, 2 Oct 2014, gordonjcp at gjcp.net wrote:

> I found Unity to be far, far quicker and easier to get around than anything that has gone before.  It took a bit of adjustment after using Gnome 2, but I can't see me ever going back.

Good, I have tried it a number of times (every release I install and try 
it) and for me it always ends up being more keystrokes to do anything more 
than use the browser. I know someone who uses the new gnome shell as well 
for music and loves it.

However, I have heard more complaints about both of them than praise. Both 
require an up to date reasonably powerful machine to run. (My P4 box will 
not even start gnome screen and unity barely moves, my 3 year old netbook 
doesn't like either of them either) As such, even something that runs 
these ok may not be able to do the same lowlatency audio that other DEs 
can do on the same machine. In trying to help people do so I have seen 
more people switch than not. Now obviously I won't have tried to help 
those for whom unity has just worked so my POV is bound to be one sided.

>> As someone who tries to get the most out of anything I use, I find most
>> commercial software extremely frustrating in the way it strait-jackets users. I
>> think this also blocks curiosity and maybe stops more youngsters joining the
>> creative communities.
> But that starts to sound like the user arguments that put me off 
> developing Linux audio software in the first place.  There seems to be a 
> mindset of "software shouldn't waste cycles looking nice and being easy 
> to use" which suggests that things ought to look shite and be difficult 
> to use with every conceivable option exposed to the user, because it 
> gives them "more flexibility".

There does need to be a balance. However, part of the problem is user 
mindset, and here I am talking linux hobby mindset. Many people who use 
linux expect to get away with a lesser machine. The same person will spend 
twice what their computer is worth for an audio IF. Really, if someone is 
serious about music, they need to have reasonable gear. But they shouldn't 
have to have a server class box just so the display can look pretty 
either. If we were truely worried about cpu cycles we would all be using 
NAMA as our DAW with no X running. One of the things I heard from one of 
the videos mentioned was that the UI can influence the music that comes 
out of it. Limitations in the UI can shape the way the music is made. So 
here is another place there needs to be balance.

I don't know how this relates, but It may be interesting to note that I 
have tried LMMS and Bitwig both of which are supposed to be very easy to 
use, but I found them hard to get noise out of at all. Yet I was able to 
use Ardour and have it act as expected from my first use. So obviously my 
head does not work the same as the average musician... and as such maybe I 
should stay as far away from developing anything as I can.

> I learned the same lesson myself after coming at it the hard way when I 
> wrote an eight-operator FM softsynth where any operator could be routed 
> to any other operator (even itself) in a giant matrix of 64 controls.

I think even the 32 setups on the DX7 were more than needed, yet there are 
people who have played with them for years who comment they are still 
learning new things on them. I do not know if two more operators who have 
helped, but having 6 still seems to be better than 4... though 4 with more 
than just sine wave in may be a different story.

> Yes, it was extremely powerful and versatile, but actually it turned out 
> that this didn't make it useful.  It needed to be *less* powerful and 
> versatile to filter out all the useless combinations.  Ever wonder why 
> your DX21 has only got eight "algorithms" by which the operators may be 
> combined?  *That's* why.

The dx21 only has 8 because it only has 4 operators. This is math related 
rather than usablity related. I suspect it had something to do with 
production cost as well. I have a 4 operator synth (even cheaper version 
than the 21) and I do not like the sound as much as the DX7 (I have one of 
those too). However the usability angle really shows well here. The DX7 
was not at all easy to program and it showed, it was easy to recognize the 
DX7 in popular music becasue everyone just used the factory defaults... 
good thing they were reasonable defaults for the DX7 would have been dead 
as soon as it came out. Having more operator controls may have helped, but 
the cost would have been a lot more too (the DX1 for example).

The manufacturers have learned from this all and most new synths just have 
presets that eveyone uses. No one creates new sounds on them but rather 
buys them for the sounds that they come with. In the end it has led to 
less experimental music, but there are a lot of other factors that pointed 
that way anyway. The big labels are not interested in music at all, just 
profit. They hire the same producers who use the same sounds that have 
worked before... A live drummer may play his own drums, but the producers 
will remove his sound and replace it with a sampled drum kit in time with 
what would have been the drumkit. One has to look for experimenting in 
music or just something different. thanks to Internet people can self 
release what the record company is not interested in. Some is awful, but 
there is a lot of really good stuff too.

> I think the design should be led by someone with experience in observing 
> what people actually do with the tools that are presented to them.  It's

Yes and no. That would be like watching some one play the guitar and 
deciding we only need the bottom 5 frets because the player never used any 
bar chords (it would be easy to find a group of 1000 professional 
musicians who played this way). The tool can define the music as already 
stated by a few people. Ease of use is not king... otherwise we would use 
the auto-harp instead of the guitar (there are some people who do amazing 
things with the auto harp, but there are a lot more that just push the 
buttons). Adding more UI than the working musician uses is not bad, it 
means that new styles can be conceived and what is never done now can 
become normal. See my comment above about different DAWs. The ones I find 
hard to use are the "Cats meow" for some of the people who have started 
their DAW experience having never recorded music with all hardware. (there 
are also some people who are just more adaptable than I am too)

I am often amazed at the new ways of making sound on physical instruments, 
some that have been around for 100s of years. It is easy to make sw more 
restricting than that. Maybe for the "average working musician" that is 
ok, it pays the bills and the asier the better. Maybe Linux developers 
should cater to these people more. On the other hand, maybe tomorrows 
closed SW will be based on tricks that the experimenter figured out using 
a more open aproach. FM synthisis was a very hard sell. No one wanted it. 
Yamaha took it only because one of their people was a tinkerer and messed 
with it enough to get it in the door. What other ideas are waiting to be 
used that will only be discovered by someone with tools that are open 
enough to allow it?

SHould Linux target those who only see a comodity? WHo are only looking to 
have what their "idol" uses? Or who want the cheapest one that works? The 
stuff already out there will be what gets bought. Developing HW with Linux 
is like developing with any other OS, it requires innovation and lots of 
support. The linux HW has to have what nothing else does and the something 
has to be seen as needed. Lets see how the mod duo does.

Len Ovens

More information about the Linux-audio-dev mailing list