Scribd
Upload
156 views

Example of Review Text

This is a review of Speech Repair Quick Error Correction Just by Using Selection Operation for Speech Input Interfaces, written by Yohana Ekky.

Uploaded by

Yohana Ekky Tan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
156 views

Example of Review Text

This is a review of Speech Repair Quick Error Correction Just by Using Selection Operation for Speech Input Interfaces, written by Yohana Ekky.

Uploaded by

Yohana Ekky Tan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

1

Speech Repair : Quick Error Correction Just by Using Selection Operation for Speech
Input Interfaces
Jun Ogata and Masataka Goto
Yohana Ekky P
C11.2011.01241

Software which is called automatic speech recognition (ASR) is created to help people
in recording data. According to William Tapscott, Yahoo Contributor Network, there are
some reasons why using speech recognition is important. First, dictation in which speech
recognition can help record data without writing on computers or gadgets that have small
keyboard that makes it faster than writing. This is very useful especially for those who are
slow typists. Second, software navigation in which speech recognition helps to carry out
commands on computers or gadgets without being troubled of clicking and maneuvering. The
last one is conversation recording in which speech recognition helps to capture a valuable
record for their use. However, using conventional speech recognition doesnt accelerate data
recording as what most users want but instead slows it down when the speech recognition
cannot recognize the voice the user inputs into it. It will be uncomfortable as well if users
need to correct errors on speech input because they have to detect, define, and correct each
error in the recognition results after speaking. In addition, since the current ASR is error
prone, there will be recognition errors left unremoved. Therefore, a user interface that allows
a user to correct (repair) recognition errors must be established.
Through this research, Jun Ogata and Masataka Goto proposed a solution to overcome
the problems occurred in speech recognition which is called Speech Repair. It functions to
help recognition errors quickly and easily corrected by users because it enables users to do
errors correction simultaneously while speaking the words.
The way speech repair works is that it automatically shows the usual word sequence as
a recognition result and also other competitive candidates without waiting an utterance ends
so that the user can correct erroneous words merely by selecting the appropriate candidate
even during the time of speaking to the recognizer. In this speech repair, there are some
features that make the errors recognition easier, immediate correction function, suspension
function and confusion network.
First, immediate correction function. It enables users to correct erroneous words right
away by selecting competitive candidate whenever the erroneous words are found.
2

Second, suspension function. It enables users to deliberately suspend and continue the
recognition process whenever they need to stop uttering to correct all errors and when they
continue uttering the rest of the sentence, the recognizer continue the recognition process
during the speech input.
Third, confusion network. It is a simple network showing the intermediate recognition
result and is received by condensing a huge word graph in order to make the users understand
when they need to find erroneous words in the recognition results.
This speech repair has been proved efficient by the experiments done using 100
sentences to be uttered by a number of speakers in Japanese Newspaper Article Sentences
(JNAS) experiment and in Corpus of Spontaneous Japanese (CSJ) experiment. The result
shows that in the JNAS experiment, recognition performance became 99.36% from only
86.70% and in the CSJ experiment 96.16% from 77.79%. It indicates that with the help of
speech repair, the word correctness was drastically improved in both evaluation data because
in both the experiments, almost all erroneous words could be corrected even when the
number of candidates displayed was ve.
In summary, speech repair is really needed in speech recognition because it works
efficiently when users find errors after uttering the words. Through the experiments done, the
effectiveness and usefulness of speech repair have been confirmed by its way in overcoming
the problem of speech recognition errors which were almost entirely corrected.
By reading this journal I assume that speech recognition which is already used in
computers and gadgets will get significant improvement. Every user will be a lot more helped
whenever using speech recognition is needed to be done without worrying about wasting time
to correct error recognition. Ogata and Goto have done great job in developing speech repair
and proved its effectiveness and usefulness by some accurate experiments. Nevertheless, they
didnt provide enough previous literature to strengthen their reason in establishing speech
repair.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy