Latest inventions in natural language tools


What is speech recognition

Speech recognition (in many contexts alsoMost speech recognition users would tend to
known as 'automatic speech recognition',agree that dictation machines can achieve
computer speech recognition or erroneously asvery high performance in controlled
Voice Recognition) is the process ofconditions. Part of the confusion mainly
converting a speech signal to a sequence ofcomes from the mixed usage of the term speech
words, by means of an algorithm implementedrecognition  and  dictation.
as  a  computer  program.
Speaker-dependent dictation systems requiring
Speech recognition applications that havea short period of training can capture
emerged over the last few years include voicecontinuous speech with a large vocabulary at
dialing (e.g., Call home), call routingnormal pace with a very high accuracy. Most
(e.g., I would like to make a collect call),commercial companies claim that recognition
simple data entry (e.g., entering a creditsoftware can achieve between 98% to 99%
card number), preparation of structuredaccuracy (getting one to two words out of one
documents (e.g., a radiology report), domotichundred wrong) if operated under optimal
appliances control and content-based spokenconditions. These optimal conditions usually
audio search (e.g. find a podcast wheremeans the test subjects have 1) matching
particular  words  were  spoken).speaker characteristics with the training
data, 2) proper speaker adaptation, and 3)
Voice recognition or speaker recognition is aclean environment (e.g. office space). (This
related process that attempts to identify theexplains why some users, especially accented,
person speaking, as opposed to what is beingmight actually find that the recognition rate
said.could be perceptually much lower than the
expected  98%  to  99%).
Speech  recognition  technology
Other, limited vocabulary, systems requiring
In terms of technology, most of the technicalno training can recognize a small number of
text books nowadays emphasize the use ofwords (for instance, the ten digits) from
Hidden Markov Model as the underlyingmost speakers. Such systems are popular for
technology. The dynamic programming approach,routing incoming phone calls to their
the neural network-based approach and thedestinations  in  large  organizations.
knowledge-based learning approach have been
studied  intensively  in the 1980s and 1990s.Both acoustic modeling and language modeling
are important studies in modern statistical
Performance  of  speech  recognition  systemsspeech recognition. In this entry, we will
focus on explaining the use of hidden Markov
The performance of a speech recognitionmodel (HMM) because notably it is very widely
systems is usually specified in terms ofused in many systems. (Language modeling has
accuracy and speed. Accuracy is measured withmany other applications such as smart
the word error rate, whereas speed iskeyboard and document classification; please
measured  with  the  real  time  factor.refer to the corresponding entries.)



1 A B C D 58 59 60 61 62 63 64 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108