Campus Event Calendar
Event Entry
New for: D1, D2, D3, D4
What and Who
Automatic Speech Recognition and Pronunciation
Catia Cucchiarini und Helmer Strik
Dept. of Language and Speech Univ. of Nijmegen
Phonetisches Kolloquium
AG 1, AG 2, AG 3, AG 4
AG Audience
Note: We use this to send email in the morning.
Date, Time and Location
Friday, 14 December 2001
14:00
-- Not specified --
17.2 - Computerlinguistik
R 5.09
Saarbrücken
Abstract
In our presentation we will talk about 4 research topics which are
all related to Automatic Speech Recognition (ASR) and Pronunciation.
1. Pronunciation Variation Modeling for ASR
The pronunciation variation that is present in speech (especially in
spontaneous, extemporaneous speech) will deteriorate the performance
of ASRs if it is not well accounted for. By modeling pronunciation
variation the performance of ASRs can be increased. I will present
an overview of our research on pronunciation variation modeling, in
which we have used both data-driven and knowledge-based methods.
2. Automatic Phonetic Transcription
For many applications phonetic transcriptions are needed. However,
making phonetic transcriptions is time consuming and thus expensive.
Since the number and size of speech corpora is growing rapidly, there
clearly is a need for automatic phonetic transcription (APT). An ASR
can be employed to make APTs. We have studied how varying the
properties of the ASR influences theresulting APTs. An important
outcome of our research is that ASRs with lower word error rates
do not always produce APTs of higher quality.
3. Phonetic transcriptions in the Spoken Dutch Corpus (CGN):
how to combine efficiency and good transcription quality
In this talk I will present the results of an experiment aimed
at establishing how phonetic transcriptions for the large CGN
corpus can be obtained most efficiently. This experiment
explores the potential of an automatically generated
transcription (AGT) by comparing an AGT with a reference
transcription (Tref) of the same material, to determine
whether and how the AGT can be improved to make it more
similar to Tref. The results indicate that the AGT can be
optimized through pronunciation variation modelling so as to
make human corrections more efficient or even superfluous, at
least for some speech styles.
4. Quantitative assessment of second language learners?
fluency: comparisons between read and spontaneous speech
In this presentation I will report on the results of two
experiments aimed at exploring the relationship between
objective properties of speech and perceived fluency in read
and spontaneous speech. The aim is to determine whether such
quantitative measures can be used to develop objective fluency
tests. Fragments of read speech (Experiment 1) of 60 non-
native speakers of Dutch and of spontaneous speech (Experiment
2) of another group of 60 non-native speakers of Dutch were
scored for fluency by human raters and were analyzed by means
of a continuous speech recognizer to calculate a number of
objective measures of speech quality known to be related to
perceived fluency. The results show that the objective
measures investigated in this study can be employed to predict
fluency ratings, but the predictive power of such measures is
stronger for read speech than for spontaneous speech.
Moreover, the adequacy of the variables to be employed appears
to be dependent on the specific type of speech material
investigated and the specific task carried out by the speaker.
-----------------------------------------------------------------
Alle Interessierten sind herzlich eingeladen.
Phonetisches Kolloquium im www:
http://www.coli.uni-sb.de/phonetik/ph-kolloquium.html
--------------------------------------------------
Stefan Baumann
University of the Saarland
FR. 4.7 Phonetics
P.O. Box 151150
D-66041 Saarbruecken
--------------------------------------------------
Contact
--email hidden
passcode not visible
logged in users only
Uwe Brahm, 04/12/2007 12:11 -- Created document.