Glossary of Terms and Acronyms

Programmers and engineers of all disciplines or nationalities love their TLAs; speech synthesis is no different. We hope to have covered all those used in this manual, and perhaps a few more. If you find any we missed (or got wrong!), please let us know for future versions...

ACL-DCI

Association for Computational Linguistics - Data Collection Initiative.

ACPA

Audio Capture and Playback Adapter.

ANN

Artificial Neural Network.

ASCII

American Symbolic Code for Information Interchange.

ASR

Automatic Speech Recognition.

ATR

Advanced Telecommunication Research.
http://www.atr.co.jp/

BEEP

British English Example Pronunciation.

btosps

Binary TO Signal Processing System.

car

Not an acronym. `Lisp' expression which refers to (and selects) the first item of a list held in a variable. See also `cdr'.

CAT

Not an acronym. CATegory of a word. HLP tags used to categorize words into Nouns, Verbs, Prepositions, etc. See NP, VP and PP.

cdr

Not an acronym. `Lisp' expression which refers to (and selects) the items of a list held in a variable, less the first item. See also `car'.

CELP

Code-book Excited Linear Prediction.

CEPLMA

CEpstral Resynthesis using a Logarithmic Moving Average filter.

CHATR

Collective Hacks from the Advanced Telecommunications Research laboratories. Well you did ask...
http://www.itl.atr.co.jp/chatr/

CMU

Carnegie Mellon University.
http://www.cs.cmu.edu/People/air/consortium/description.html

CSTR

Centre for Speech Technology Research. A department of Edinburgh University, UK.
http://www.cstr.ed.ac.uk/

CSLU

Center for Spoken Language Understanding. A department of Oregon Graduate Institute of Science and Technology, USA.
http://www.cse.ogi.edu/CSLU/

CVS

Concurrent Versions System. CVS is a front end to the RCS revision control system. It extends the notion of revision control from a collection of files in a single directory, to a hierarchical collection of directories consisting of revision controlled files. These directories and files can be combined together to form a software release. CVS provides the functions necessary to manage these software releases and to control the concurrent editing of source files among multiple software developers. CVS keeps a single copy of the master sources. This copy is called the source `repository'; it contains all the information to permit extracting previous software releases at any time based on either a symbolic revision tag, or a date in the past.

darpa

Defense Advanced Research Projects Administration. The central research and development organization for the Department of Defense (DoD), USA.
http://www.darpa.mil/

DTW

Dynamic Time Warping.

EGG

Electro-Glottal Graph. Device for measuring throat movement caused by speaking.

EMACS

Editor MACroS. A Macro-based editor and complete computing task environment.

ESPS

Entropic Signal Processing System.

FSF

Free Software Foundation.
http://www.gnu.ai.mit.edu/fsf/

HLCB

High Low Continuation Boundary. Tags used to mark intonation on syllables.

HLP

High Level Phrasing. Method of tagging speech with prosodic information.

HMM

Hidden Markov Model.

Holmes

John Holmes, one of the founders of speech synthesis.

HTK

Hidden (Markov model) Tool Kit. A product of Entropic Research Laboratory, Inc.
http://www.entropic.com

IFT

Illocutionary Force Type. Strength or emphasis put on a phrase. Speech act information - meaning you want to convey above and beyond just the words spoken. As an example, the English phrase `I understand' can mean `Thank you for informing me (I'm happy)' or `Now I know what you intend I'm not happy' or even `I heard what you said but haven't a clue what you mean' depending on how and when it's said. That's IFT at work. The simplest case is the difference between a question and a statement using the same words.

IntoneStream

Series of symbols representing the intonation required on an utterance. Attached to the WordStream.

IPA

International Phonetic Association. Representative organization for phoneticians.
http://www.arts.gla.ac.uk/IPA/ipa.html

JToBI

Japanese Tones and Break Indices.

jtts

Japanese Text-To-Speech.

LDC

Linguistic Data Consortium. A group established to broaden the collection and distribution of speech and natural language databases for the purposes of research and technology development in automatic speech recognition, natural language processing and other areas where large amounts of linguistic data are needed.
http://www.ri.cmu.edu/comp.speech/Section1/Data/ldc.html

LFG

Lexical Functional Grammar.

LISP

LISt Processing language. A programming language originally developed for Artificial Intelligence (AI) but now used mainly in the speech synthesis field.

LMA

Logarithmic Moving Average. Mathematical reference to a method used in audio filtering. See CEPLMA.

LPC

Linear Predictive Coding.

LTS

Letter To Sound.

LVQ

Learned Vector Quantization.

M-ACPA

Multimedia - Audio Capture Playback Adapter.

MARSEC

MAchine-Readable Spoken English speech Corpus.

MFCC

Mel Feature Cepstral Co-efficients.

mtts

Multi-lingual Text-To-Speech.

mrpa

Machine Readable Phonetic Alphabet.

Mu-law

Not an acronym. Pronounced `mew-LAW' - the `Mu' is actually the Greek letter `Mu'. An 8-Bit compression code for audio signals including speech. It is widely used in the telecommunications field because it improves the signal-to-noise ratio without increasing the amount of data. It is a companding technique. That means it carries more information about the smaller signals than the larger. Sometimes appears in documents written as `ULAW'.

MULE

MUlti Language Editor. Extended part of EMACS.

NFS

Network File System. A distributed file system that provides transparent access to files residing on remote disks. Developed at Sun Microsystems in the early 1980's.

NIST

(American) National Institute STandards.

NLP

Natural Language Processing.

NN

Neural Network.

PN

Noun Phrase. HLP tag used to denote an input word as a Noun.

nus

Non-Uniform (unit) Selection.

nuuph

Not an acronym. The `nuu' is the Greek letter `Nuu'. Japanese phoneme set.

NUUCEP

Not an acronym. The `NUU' is the Greek letter `Nuu'. NUUtalk CEPstral synthesis routines.

OAPD

Oxford Acoustic Phonetic Database. Contains data on vowel-consonant and consonant-vowel combinations in both stressed and unstressed locations.

PhoneStream

Series of symbols representing the phonemes of an utterance. Attached to the WordStream.

PhonoWord

Type of input accepted by CHATR. Allows specification of prosodic phrases and intonation features. Utterance is tagged with four letters (D=Discourse, S=Sentence, C=Clause and P=Phrase) to specify phrase levels, and other letters (e.g. H and L) to indicate emphasis and accent.

PP

Preposition Phrase. HLP tag used to denote an input word as a Preposition.

PphraseStream

Series of symbols representing the prosodic phrases of an utterance. Attached to the WordStream.

PSOLA

Pitch Synchronous Over-Lap and Add. Algorithm to independently modify the fundamental frequency and duration of a speech signal. Used during concatenation of selected units from a finite speech database such that minimal prosodic damage occurs due to target/selected unit mismatch.

RCS

Revision Control System. A system that keeps track of different versions of files. If one person is editing a source no other developer may do so. Thus all sources are by default read-only. When a file is checked out by a developer, they may change it but no other developer may check it out at the same time. When a developer is finished, they may check in the file thus allowing others to check it out.

RFC

Rise Fall Continuation. A now become dated method of tagging phoneme-sized segments with duration and frequency values.

SegStream

Series of symbols representing the segments of an utterance. Attached to the WordStream.

SGML

Standard Generalized Markup Language.

SphraseStream

Series of symbols representing the syntactic phrases of an utterance. Attached to the WordStream.

Stream

One of a sequence of cells containing symbols generated and/or interpreted by CHATR and linked to an utterance (and other streams). Causes changes in the timing, intonation and prosody of the synthesized output.

SylStream

Series of symbols representing the syllables of an utterance. Attached to the WordStream.

TIMIT

A large speech corpus from TI and MIT.

TLA

Three (or sometimes more or less) Letter Acronym. Initials represent a well (or often un)-known title or description.

ToBI

Tones and Break Indices.

tts

Text-To-Speech.

ULAW

Not an acronym. Pronounced `mew-LAW' - the `U' is actually the Greek letter `Mu'. An 8-Bit compression code for audio signals including speech. It is widely used in the telecommunications field because it improves the signal-to-noise ratio without increasing the amount of data. It is a companding technique. That means it carries more information about the smaller signals than the larger. Sometimes appears in documents written as `Mu-law'

utterance

A series of words you wish CHATR to synthesize as speech. Basically the input to CHATR, in whichever form it may take.

VP

Verb Phrase. HLP tag used to denote an input word as a Verb.

VQ

Vector Quantization.

WordStream

Series of words to be `spoken' by CHATR, derived from the utterance.

XMG

X Multi-Graph. A graphics display program written at CSTR, Edinburgh University, UK.
http://www.cstr.ed.ac.uk/

XWAVES

Not an acronym. A graphics display program from Entropic Research Laboratory, Inc.
http://www.entropic.com

Go to the first, previous, next, last section, table of contents.