504 INDEX
overelaboration, 278
overgeneralization, 278
P
parallel monolingual components, 444
Parallel Phone Recognition followed by
Language Modeling (PPRLM ), 260
partial parsing, 354
part-of-speech (POS) tagging, 336, 410
parts-of-speech (POS), 354
PCA (principal components analysis), 264
PDTS (Polyphone Decision Tree
Specialization), 112, 295
peculiarities of language, 83–90
overview, 83
prosody and phonology, 86–87
segmentation and morphology, 87–89
semantics and pragmatics, 89–90
sound systems, 85–86
written script and relation to spoken
speech, 83–85
perception, 277
perceptual benchmarks, human language
identification of, 236–237
perceptual linear prediction (PLP), 249
phonemes, 12, 92
phones, 12
phone sets and acoustic modeling, 162–164
phonetic confusion, non-native speech
characteristics of, 280–282
phonetic context questions, 109
phonetic coverage across languages, 99–102
phonetics, 12–16
phonetic sound inventories, 92–93
phonographic systems, 23
phonology, 12–16, 86–87
phonotactic language identification, system
architectures in, 261–262
phonotactic modeling, 244, 251–262
basic N-gram modeling, 252–253
extended phonotactic context, 255–259
context clustering with
maximum-likelihood binary
decision trees, 256–259
overview, 255–256
N-Gram estimation and smoothing, 253–254
overview, 251–252
system architectures in phonotactic LID,
261–262
phonotactics, 14, 244
phrase alignment models, statistical direct
translation of, 325–327
phrase correspondence score, 354
pidgins, 11
PLP (perceptual linear prediction), 249
polynomial classifier, 246
polyphone, 75
Polyphone Decision Tree Specialization
(PDTS), 112, 295
polyphoneme, 92
portable speech-to-speech translation, see AT R
system of portable speech-to-speech
translation
POS (part-of-speech) tagging, 336, 410
POS (parts-of-speech), 354
power distance, 311
PPRLM (Parallel Phone Recognition followed
by Language Modeling), 260
pragmatics and semantics, 89–90
precision, 189
prefabrication, 278
principal components analysis (PCA), 264
probabilistic language identification
framework, 242–245
probabilistic output, 306
probabilistic systems, 305
pronunciation adaptation, 276, 293–295
pronunciation dictionary resources, 157
pronunciations, canonical, see canonical
pronunciations
pronunciation variants, 158–162
prosodic language identification, 262–265
prosodic modeling for multilingual speech
synthesis, 216–219
prosodic models, 210, 245
prosodic parameters, generation of, 369
prosody, 12–16, 86–87
prototype S2ST system, 392
psycholinguistic study using speech
resynthesis, 239–240
Q
QTr (Quick Transcription), 37
quantitative feedback, 306
quantitative systems, 305

Get Multilingual Speech Processing now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.