logo

MATE Deliverable D1.1

Supported Coding Schemes

 

INTSINT

 

INTSINT is a coding system of intonation developed by Daniel Hirst and his colleagues at the CNRS centre of the Aix-en-Provence University. Descriptions of this method can be found in [Hirst 91,94]; [Hirst et al. ?].

More information about the INSTINT system can be found at http://www.lpl.univ-aix.fr/~hirst/intsint.html.

Coding book:

There is no coding book to apply INTSINT, although the tutorial describing the use of the automatic annotation tool includes some references about the philosophy of INTSINT.

Applications:

[Llisterri 96a]

"The system has already been applied to several languages (see, for example, [Hirst et al. 93] and is being used in MULTEXT Multilingual Text Tools and Corpora project ([Hirst et al. 94]; more information on the project is available at URL http://www.lpl.univ-aix.fr/projects/multext/index.html) for the encoding of intonation in the paragraphs contained in the EUROM.1 corpus."

Within the MULTEXT project, three different groups have been working in the annotation of read paragraphs extracted from the EUROM.1 corpus:

a) Laboratoire Parole et Langage, CNRS, Aix-en-Provence, France

b) Grup de Fonètica, Departament de Filologia Espanyola, Universitat Autònoma de Barcelona, Spain

c) Department of Phonetics, Umeå University, Sweden

These three groups have been working in the annotation of English (40 passages), Spanish (40 passages), French (40 passages), German (20 passages) and Swedish (5 passages).

In all cases, the annotated material were read-aloud paragraphs. No real dialogues were annotated.

Evaluation:

There is no formal evaluation of INTSINT, but an evaluation of the performance of the annotation tool using INTSINT ('mes') has been carried out within the MULTEXT project. The results of this evaluation are presented in [Llisterri 96b].

Purpose and underlying approach:

INTSINT is based of the intonation model developed by Daniel Hirst and the group of the Aix-en-Provence University [Hirst 91, 94], [Hirst et al. ?].

It is conceived "to provide a purely formal encoding of the macroprosodic curve. Each target point of the stylized curve is coded by a symbol either as an absolute tone, defined globally with respect to the speakers pitch-range or as a relative tone, defined locally with respect to the immediately neighbouring target-points" ([Campione et al. 97], p. 72).

List of phenomena annotated:

INTSINT includes only symbols to transcribe F0 events:

1) Absolute Tones

INTSINT includes three symbols to label the Absolute Tones, which are defined according to the speaker's pitch range.

T

top of the speaker's pitch range

M

initial, mid value

B

bottom of the speaker's pitch range

2) Relative tones

Relative tones are coded in INTSINT considering the height of the preceding and following target points. Five different symbols exist to transcribe these Relative Tones:

H

target higher than both immediate neighbours

L

target lower than both immediate neighbours

S

target not different to preceding target

U

target in a rising sequence

D

target in a falling sequence

Examples:

The INTSINT coding is usually stored in a set of files generated with the annotation tool 'mes', which is described below. An example of the files containing the INTSINT coding of the speech utterance 'Il faut que je sois a Grenoble Samedi vers quinze heures' is provided here:

Orthographic representation:

il 1745

faut 3605

que 6669

je 8710

sois 10678

a 14461

Grenoble 15334

Samedi 26463

vers 37025

quinze 39730

heures 45645

# 50164

 

F0 target points:

-99.95 140.012

106.3 163.589

265.6 217.241

521.7 148.048

617.25 190.525

827.95 130.806

1249.35 223.515

1614 139.595

1822.45 172.134

1983.25 144.184

2078.75 185.903

2248.35 152.624

2505.85 99.1518

2730 152.606

 

INTSINT transcription:

M -1999

L 2126

T 5312

M 10434

H 12345

L 16559

T 24987

M 32280

H 36449

L 39665

H 41575

D 44967

B 50117

M 54600

 

The following picture shows the visualization of transcription in the environment 'mes' (see below).

Markup language:

Symbolic labels time-aligned with the signal. Numerical values for f0 target points.

Annotation tools:

A tool for prosodic annotation using INTSINT (see picture above) has been developed in the framework of the MULTEXT project: 'mes' allows automatic annotation of speech signals from phonetic data (a stylized version of the F0 contour obtained by the MOMEL automatic stylization procedure).

A description of this tool is available at:

http://www.lpl.univ-aix.fr/valorisation/mes_signaix/