SynPaFlex Corpus Annotation Manual PDF

La Venus dIlle

Sans doute ils sont incrustés dans le bronze. Ce sera peut-être quelque statue romaine. — Romaine ! c'est cela. M. de Peyrehorade dit que c'est

Characterisation and generation of expressivity in function of

03-Feb-2021 styles de parole pour la construction de livres audio. Thèse présentée et soutenue à Lannion le 02 ... La Vénus d'Ille

SynPaFlex Corpus Annotation Manual

04-Oct-2019 Global particularities of the speech audio files and their texts were written for each book

Building a Search Engine for Music and Audio on the World Wide

tiques l't points d'iutc>r('t concernant la nature des fichiers audio sur le veb. Shirk~". alld Linda Kuopke have always been there for Ille.

français langue étrangère

14-Feb-2019 Cahier d'exercices + Livre numérique de l'élève. • Guide pédagogique avec Tests et ressources + 2 CD audio pour la classe. • Livre numérique de ...

FRW 3101: Introduction à la littérature française II Dr. Rori Bloom

la littérature à enchanter à travers des représentations du magique du surnaturel ou du Mérimée

Caractérisation et génération de lexpressivité en fonction des styles

styles de parole pour la construction de livres audio. Thèse présentée et soutenue à Lannion le 02 Octobre 2020 La Vénus d'Ille

Jocastas Divine Head: English with a Foreign Accent - DS CARNE

Venus makes love to Mars which prompted a notable essay of. Montaigne: Elle avait dit et

A complete treatise on French pronunciation

Voila le bureau d'acajou. Made-. 24. When does the syllabie dirision take place in words composed of compound sounds?

SynPaFlex Corpus Annotation Manual

Aghilas Sini

1, Damien Lolive1,´Elisabeth Delais-roussarie2,

Marie Tahon

1, and Ga¨elle Vidal1

1 IRISA, University of Rennes 1 (ENSSAT), Lannion, France

2UMR 7110-LLF & Universit´e Paris-Diderot, France

October 4, 2019

1 Corpus files2

1.1 Audio files

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.1.1 Global description

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.1.2 Later changes made to the audio files

. . . . . . . . . . . . . . . . . . . . . 4

1.2 Text files

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.3 Manual annotations

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.3.1 Characters annotation process

. . . . . . . . . . . . . . . . . . . . . . . . . 6

1.3.2 Prosody annotation process

. . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.3.3 Emotion annotation process

. . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.3.4 Other events

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.4 Manual validation of the automatic phonetic segmentation

. . . . . . . . . . . . . . 9

2 Analysis10

2.1 Estimation of amounts of narration and direct style for each book reading

. . . . . . 10

2.2 Narration speech subjective classification

. . . . . . . . . . . . . . . . . . . . . . . . 11

2.3 Characters annotation analysis

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.4 Prosody annotation analysis

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.5 Emotion annotation analysis

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1

1 Corpus files

1.1 Audio files

Global particularities of the speech audio files and their texts were written, for each book, in a corpus

description file, including an estimation from the text of amount of acted direct speech, and references

to manually produced information.

1.1.1 Global description

The SynPaFlex corpus is comprised of dozens of hours of speech recordings found on the free domain

server Librivox. Files formatted with the best quality, mp3128-kbps, can be download at:a rchive.org.

It is monospeaker, performed by a female reader who reads French classic literature (written in the

19th century for most of them). Each file corresponds to the recording of a book chapter, or of a

small literary work (poem, tale, fable, ...). The speaker can personify her voice for some characters

in direct style. Only files of good audio quality were selected 1. Table 1 p roposesa litera ryge nreclassification and sho ws,f oreach genre, dura tionfo rwhole and best audio quality speech as well as for different manual processing. Chapters distributed over five historicnovels constitute the majority of the corpus:70% of the 87 hours.fantasticand adventureones represent 15% and 10% (13 and 9 hours) of the corpus. A group oftalesfrom different authors, and a few chapters of asymbolistshort story concern for each one to two hours of

recording. The rest of the speech corresponds to few minutes readings of other original literary works:

anepistolarynovel, afantastic epicpoetic prose, afantastic dramatic, apamphlet, and also verse forms offablesandpoems.SynPaflex corpus

French mono-speakerWhole corpus

audio-files durationManual annotation subcorpora audio-files durationGenresWhole corpusBest

QualityCharactersEmotion

and

ProsodyPhonetic

segmentation validationHistoric novels61h 18m57h 21m24h 43m8h 41m20m Fantastic novels and short stories12h 55m10h 08m1h 081h 08m- Adventure novels and short stories9h 01m3h 37m3h 37m3h 37m20m

Symbolist short story1h 42m1h 18m1h 42m-

Tales1h 06m1h 06m1h 06m10m-

Epistolary novel20m20m20m--

Fables18m13m18m-1m

Fantastic epic18m11m18m--

Fantastic dramatic11m10m10m-3m

Poems12m12m12m-1m

Pamphlet6m-6m--

TOTAL87h 28m74h 35m33h 34m13h 36m47m

Table 1: Presentation of the audio literary corpus and manual annotation 1 Unselected recordings remain available, for example the 12 hours novelLes temps modernes(link). 2

GenreTitle, author(full reading *)DurationLower

audioqualityAbbr.

Historic novelsLes Myst

`eres de Parisvol.1 and 2"Eug`ene Sue*25h 32m22mHNmy

Les mis

´erables, Victor Hugo14h 16m2h 18mHNmi

Madame Bovary, Gustave Flaubert *13h 12m26mHNma

Le roman de la momie,Th´eophile Gautier*6h 33m-HNro

Germinal,´Emile Zola1h 46m50mHNge

Fantastic novels

and short storiesLa vampire, Paul F´eval *10h 02m2h 26mFNva Voyage au centre de la terre, Jules Verne1h 52m22mFNvo La V ´enus d"Ille, Prosper M´erim´ee *1h 02m-FSve

Adventure novel

and short storiesLa fille du pirate, Maurice Chevalier *6h 43m4h 31mANfi

Carmen, Prosper M´erim´ee *2h 18m53mASca

Symbolism

short storiesContes cruels,Auguste Villiers de l"Isle-Adam1h 42m24mSYco

TalesLa malle volante, Andersen *12m-TAan

Le monstre Yatama, Claudius Ferrand *8m-TAfx

Les sept chevreaux, Claudius Ferrand *16m-TAfy

Ourashima Taro et la d

´eesse de l"Oc´ean,Claudius Ferrand*16m-TAfz La Hy `ene, l"Hippopotame et l"´El´ephant,Franz de Zeltner*11m-TAzxquotesdbs_dbs2.pdfusesText_2

[PDF] SynPaFlex Corpus Annotation Manual

SynPaFlex Corpus Annotation Manual

Aghilas Sini

1, Damien Lolive1,´Elisabeth Delais-roussarie2,

Marie Tahon

1, and Ga¨elle Vidal1

2UMR 7110-LLF & Universit´e Paris-Diderot, France

October 4, 2019

Contents

1 Corpus files2

1.1 Audio files

1.1.1 Global description

1.1.2 Later changes made to the audio files

1.2 Text files

1.3 Manual annotations

1.3.1 Characters annotation process

1.3.2 Prosody annotation process

1.3.3 Emotion annotation process

1.3.4 Other events

1.4 Manual validation of the automatic phonetic segmentation

2 Analysis10

2.1 Estimation of amounts of narration and direct style for each book reading

2.2 Narration speech subjective classification

2.3 Characters annotation analysis

2.4 Prosody annotation analysis

2.5 Emotion annotation analysis

1 Corpus files

1.1 Audio files

1.1.1 Global description

19th century for most of them). Each file corresponds to the recording of a book chapter, or of a

French mono-speakerWhole corpus

QualityCharactersEmotion

ProsodyPhonetic

Symbolist short story1h 42m1h 18m1h 42m-

Tales1h 06m1h 06m1h 06m10m-

Epistolary novel20m20m20m--

Fables18m13m18m-1m

Fantastic epic18m11m18m--

Fantastic dramatic11m10m10m-3m

Poems12m12m12m-1m

Pamphlet6m-6m--

TOTAL87h 28m74h 35m33h 34m13h 36m47m

GenreTitle, author(full reading *)DurationLower

Historic novelsLes Myst

Les mis

´erables, Victor Hugo14h 16m2h 18mHNmi

Madame Bovary, Gustave Flaubert *13h 12m26mHNma

Germinal,´Emile Zola1h 46m50mHNge

Fantastic novels

Adventure novel

Carmen, Prosper M´erim´ee *2h 18m53mASca

Symbolism

TalesLa malle volante, Andersen *12m-TAan

Le monstre Yatama, Claudius Ferrand *8m-TAfx

Les sept chevreaux, Claudius Ferrand *16m-TAfy

Ourashima Taro et la d