![]() |
14th International Congress of Phonetic Sciences (ICPhS-14)San Francisco, CA, USA |
![]() |
This paper aims to contribute to the debate about autonomous vs
interactive theories of spoken speech recognition supporting the
latter.
We will present the results of three recognition
experiments based on two monologues produced by Italian
talkers.
In the first experiment shadowing technique was used in order
to investigate the amount of intelligibility of the speech chain:
the listener was asked to repeat with the shortest delay what
he/she heard.
In the second experiment the continuity of the monologues
was interrupted by inserting a pause after each word.
In the third experiment only lexical words in a randomised order
were presented, a silence was inserted between words.
In the last two experiments listeners were asked to rapidly
repeat heard words, missed recognition were counted.
Results show that the higher level of intelligibility derives from
natural listening condition realised in exp. 1, recognition
percentages decrease from the first to the last experiment.
Bibliographic reference. Albano Leoni, Federico / Cutugno, Francesco (1999): "The role of context in spontaneous speech recognition", In ICPhS-14, 861-864.