background image
ost YLE radio programmes today
are created in various CAR (com-
puter-assisted radio) systems that integrate
planning, text and sound editing, transmis-
sion, and reporting to various collecting
societies. If a programme is going to be
preserved in the archive after transmission,
ideally it should be possible to move the
sound files and the accompanying metadata
online to the archival system.
n theory, this should be simple, as
everything is already in a digital format.
In practice, this proved to be the most dif-
ficult part of the archival system due to
differences between and changes in the
CAR systems.
owever, parts of the system already
function and the general principle is
simple.The decision to archive a pro-
gramme is made by its producer. After the
programme has been transmitted, the pro-
ducer commands the CAR system to send
it to the archive. Metadata such as title,
content description, duration, and names
of participating individuals are converted
into XML files and sent to the catalogue
database. At the same time, the sound file is
sent to the digital mass storage.The archive
system links the sound file to the metadata,
creates a bit-reduced pre-listening version
of the sound file and tells the BRS
SEARCH user interface that the pro-
gramme can now be found in the archive.
When the system is complete, the archive
will monitor all incoming programmes on
a daily basis, and the accompanying meta-
data will be checked.Typing errors will be
corrected, personal names will be subject
to authority control, and subject headings
added. It is also expected that archivists
will give feedback to producers on the
quality of metadata received.
oday,YLE transmits about 200,000
hours of radio programmes every
feature which allows users to listen to the
complete programme and mark those parts
that they need. Instead of ordering a large
file consisting of, say, half an hour of linear
sound, the user gets only the three-minute
part that he wants.
- Access through search and retrieval system.
- User can search digital and/or analogue
- Browsing-quality sound streamed to all
users production-quality files take longer.
- Format conversion on demand.
- Special features: pre-selection of
excerpts, support for multi-part works.
he metadata management system
contains descriptive information on
the material (title, authors, performers,
duration etc.) and also technical metadata.
Each item is assigned a unique ID that is
used when creating names for individual
audio files, then, during the last stage of
the process, the audio files are imported to
the digital audio archive system and linked
to the related metadata.
ll audio files that are stored in the
archive are controlled by an essence
management system which processes the
files and also optimises access time by
caching the most recently stored or
retrieved files within the online storage. All
pre-listening copies are kept online for fast
access time.
he search and retrieval system has a
database structure, which can grant
access rights for different user groups: text
data search (open to all users), access to
pre-listening quality material (open to a
restricted group of users) and access to
production quality material (open to a
restricted group of users).
of broadcasts off-air. Neither was it
designed to support on-air broadcasting.
Before archival materials could go on air,
they would first have to be moved to a
CAR system. It was also decided that the
archive would not be used for temporary
storage. All materials that go into the
archive will, in principle, stay there forever.
The link to the existing catalogue meant
that the system does not support the digi-
tisation of uncatalogued material. It cannot
at present at least be accessed outside
our Wide Area Network.
rom the user's viewpoint, the digital
archive is not much different from the
old catalogue database, which already has
more than a thousand users within our
company.The users approach the archive
through BRS-SEARCH, a Web-browser
based user interface.They can search all of
the items in the database, or digitised
material only.
e also had to consider files with
complex relationships. How, for
instance, should we handle multi-part
musical works? In classical music, sym-
phonies and sonatas typically consist of
three or four parts. Should we handle
sonatas as one sound file, or as three or
four sound files? We chose a solution that
allows us to use both approaches: the user
can retrieve the whole work or one part
only.The same principle is applied to news
broadcasts which are treated hierarchically
as `multi-part works', allowing the user to
retrieve the whole programme or just
one story.
n principle, it would also be possible to
index other radio programmes in great
detail, so that the user could go directly to
a specific part of the programme. However,
this would demand a detailed documenta-
tion of the contents.We do not have the
capacity for this. Instead we have built a