Linking timecoded audio to other resources

[David Nathan]

Hi Pascale

Re your queries about linking timecoded audio to other resources.

This is surely one of the most important representational and research aven=
ues for linguists, but as you can see the implementation is not entirely us=
er-friendly yet. It was great to launch the question and Nick thanks for th=
e reference to Annodex.

Pascale it might help in this early stage to distinguish the following:
- (a) the kinds of knowledge/linguistic structures that you want to delinea=
te and link
- (b) the method(s) of representing (a)
- (c) the system/software/tools you use to actually construct (b), ie to au=
thor the linked data
- (d) the various audiences and usages you propose for your linked data
- (e) ways of delivering the results of (c) given (d)

In this way, you might find that there are simpler steps than jumping into =
particular technologies or dependencies upon them, or at least you can iden=
tify ways that you can decide which ones are appropriate. For example, the =
idea of using a streaming server is a jump direct to (e), without carefully=
 considering (or deciding) the intermediate steps, and, as you wrote, may n=
ot provide a key function of distribution control to the level you=FF=FFd w=
ant.
So there are 3 issues to consider:
1. For robust, long-living, shareable data, you=FF=FFd want to look careful=
ly at (a) and (b) first. The choice/design of (a) is significantly part of =
your own linguistic emphasis, although it would pay to look at, for example=
, Stephen Bird et al=FF=FFs paper on annotation schemes to see what kinds o=
f linguistic elements and structures they mention.
2. If you choose a good representational scheme, it will be robust, at leas=
t partially supported by software (and with help from others such as ELAR),=
 and will be easily transformable into other formats later. The obvious one=
 is (a suitable schema/language of) XML, although there are others such as =
Quicktime=FF=FFs QT Text (Nick Thieberger also has experience with this as =
well), and more importantly SMIL (based on XML); in addition other standard=
 representational means such as relational databases which should not be ov=
erlooked.
3. (Leaving aside the authoring for now) There is a variety of ways that th=
e authored materials could be delivered - via SMIL, Quicktime, and various =
other players that can be simply run on standalone computers, and delivered=
 either by web or CD, as well as web-servers. I do not believe that it is a=
n obvious or default choice at this stage to put raw and perhaps restricted=
 linguistic data on the web - CD/DVD meets the needs of both distribution c=
ontrol and bandwidth. Although this may well change in the future, it is be=
tter for you to make informed decisions about (a) and (b) above, and put yo=
ur emphasis in those areas, in which case any of the possibilities will be =
open to you as the technology evolves and/or the role of your data changes.

I hope that this is helpful - any follow ups appreciated.

best

David