[linux-audio-user] The Open Loop Library, a few questions

[Darren Landrum]

On Friday, December 20, 2002, at 07:00 AM, matthew yee-king wrote:

> you could use the same categories for instruments that you have in your
> samples section...

Thank you very much for your reply.

I think the instrument or instruments used in a loop would need to be 
more specific. You can't just say that the loop was made with a 
woodwind. You would need to specify that it was a recorder, or a 
piccolo, or an oboe.

However, this does give a good place to start. This data field could be 
hierarchical, where a person selects the general category (like 
woodwind), then types in a specific instrument. That brings with it its 
own problem, though: someone can select woodwind then type "marimba", 
because they accidently selected the wrong category.

> I think that might be confusing - maybe one shot samples could go into
> the samples section? drum fills will generally loop - many great drum
> tracks have been programmed entirely from drum fills :) . What i'm
> trying to say is that its a fine line between a drum fill and a drum
> beat, so fills could go in loops.

Good point. I thought about this a little more, and I think the person 
using this Acid clone will be smart enough to know that you don't 
actually have to loop a loop. :)

> Maybe the database could store a checksum of the file and link the
> metadata to this. This could work in a similar way to the cddb - you
> could look up the data for your file. I can help with the database if
> you want ...

This is cool. I have an offer to help already. :) Thank you.

There was a previous email on how Ogg can store a lot of different 
metadata as key/value pairs. That would be the ideal solution, but if 
it doesn't pan out for some reason, checksums would be a good way to go.

And of course, this brings us to more discussion on the client for this 
system, which seems to me to be the killer part of the entire app. 
(Don't worry, I still plan to make the system usable only with a 
browser.)

It seems to me that the easy way to embed metadata into each loop would 
be with the client (in fact, the client can even act as a front end to 
the Vorbis/FLAC encoder). When a person makes a loop, they would load 
it into the client, add the data via a form, and then the client can 
automatically upload it to the appropriate place.

Which brings up the question on how to do all of the same things 
without a client.

It seems to me that a person could upload a file, and the server can 
check it to see if there's any metadata. If not, then it could put the 
file in a holding area and forward the user to a form where they can 
enter the data. The server could then embed the data, and shuffle it 
off to the appropriate mirror.

This is going to be *a lot* of work. There goes all of my free time. :)

Regards,
Darren Landrum