Thematic Domain Group: Audio

The Audio TDG investigates and stimulates research on extensions to the existing ISOCat labelling scheme with the specific purpose of labelling audio data.

The TDG investigates and proposes annotation schemes in terms of organisation of tiers and sub-tiers, labels of tiers, content of tiers and the open or closed item sets to be used. Examples are label sets for foreground speakers and background speakers, phonetic labelling, alphabets and so on.

Audio data are or become available in various levels of sound quality. At one side of the quality range we have for example high quality radio or television broadcasts. For this type of audio, annotation issues are related to the way in which the information is presented: such as simultaneous mix of speech and music, overlapping speech by different speakers, jingles, background music and so on. At the other side of the quality range we have the poor-quality recordings of unique data, such as on cassette tapes used in the eighties to record the remaining speakers of endangered languages, historic noisy recordings, and similar material.

At the same time, more and more audio data will become available – a database of 10000 hours is not at all exceptional nowadays. A proper annotation may help to rapidly access these data by adequate text-based queries.

Therefore there are two points of special interest.

(1) Annotation (schemes) that can be derived from the (semi)-automatic processing of the audio signal, by e.g. procedures based on signal analyses and speech decoding.

(2) Annotation (schemes) in which verbal word-level annotation is –for whatever reason– not an immediate or trivial option, for example because the language is unknown. This is of interest in anthropological data. To that end other levels for annotation schemes must been included e.g. annotation on sub-word level. This can e.g. be envisaged by

(a) a symbol string (phone-like), possibly with errors in transcription

(b) a feature matrix with real numbers indicating ‘phonetic’ feature values

Proposals in this direction are now being taken into consideration.

The TDG strives toward a coherent and flexible use of annotation schemes across languages.

Viewswitch to the TDG specific ISOcat view
(Note: this will reload the ISOcat UI in your browser and might require you to login again!)
ChairWright, Sue Ellen (Kent State University)
MemberPrawiro-Atmodjo, Peia (CLST)
Memberten Bosch, Louis (Radboud Univ Nijmegen)