Australian National Database of Spoken Language (ANDOSL)
What if your data needs are not met by the current holdings?
We realise that the initial ANDOSL products, although substantial,
will not satisfy every need. There are however several ways in which the
existence of the ANDOSL corpus can be used to assist you.
Although the first release of the data has been designed to meet the needs
of the speech scientist or technologist who wishes to apply further computational
processing to model the data in some way, it is entirely possible for the
data to be released in other forms if the demand is there and the costs
can be met. CDROM audio tracks or even audio cassette could be considered.
You can apply jointly with the ANDOSL developers for funding to enhance
the ANDOSL holdings in a way that is very specific to your needs. Collaborative
research and research infrastructure development is very popular with granting
bodies and the fact that your data would then be available to the whole
community in a form that is technically compatible with a very large body
of related data could be very attractive. Certain researchers have already
expressed an interest in adding Developmental Speech data, more accented
forms of Australian English, Pathological Speech data, regional variants
of Australian English, and a collection of Australian Proper Names. Research
funding committees at ANU, Sydney University, and Flinders University have
already responded positively to this approach and other discussions are
Putting your existing data into an ANDOSL compatible form may enhance your
ability to collaborate with colleagues on issues of mutual concern. If
you have data which you believe may be significant for our national resource,
you could offer it to the ANDOSL project which will examine it and assess
whether we have the resources to transform your data into a form that is
technically compatible with the ANDOSL standards. It should be noted that
the ANDOSL standards relate mostly to standards of data description. High
levels of signal quality, although beneficial, are not essential for inclusion
if the data meets other criteria of usefulness. One donation of data suitable
for speech synthesis research has already been made.
Last modified: 24 March 1999