SEA_AP: a segmentation and labelling tool for prosodic analysis

Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 223# Received!22!February!2016.! Accepted!3!May!2016.! ! ! ! ! SEA_AP:!A!SEGMENTATION!AND!LABELLING!TOOL!! FOR!PROSODIC!ANALYSIS!! Paula#LÓPEZ#OTERO,#Laura#DOCÍO#FERNÁNDEZ,#Carmen#GARCÍA#MATEO,#Marta#MARTÍNEZ# MAQUIEIRA,#Rocío#VARELA#FERNÁNDEZ#&#Elisa#FERNÁNDEZ#REI*# AtlantTIC#Research#Center#for#Information#and#Communication#Technologies,# Universidade#de#Vigo#/#Instituto#da#Lingua#Galega,#Universidade#de#Santiago#de# Compostela*# [email protected]#/#[email protected]#/#[email protected]#/# [email protected]#/#[email protected]#/#[email protected]# # Abstract# This#paper#introduces#a#tool#that#performs#segmentation#and#labelling#of#sound#chains#in#phono# units,# syllables# and/or# words# departing# from# a# sound# signal# and# its# corresponding# orthographic# transcription.#In#addition,#it#also#integrates#acoustic#analysis#scripts#applied#to#the#Praat#programme#with# the#aim#of#reducing#the#time#spent#on#tasks#related#to#analysis,#correction,#smoothing#and#generation#of# graphics#of#the#melodic#curve.#The#tool#is#implemented#for#Galician,#Spanish#and#Brazilian#Portuguese.# Our#goal#is#to#contribute,#by#means#of#this#application,#to#automatize#some#of#the#tasks#of#segmentation,# labelling# and# prosodic# analysis,# since# these# tasks# require# a# large# investment# of# time# and# human# resources.#! # Keywords# segmentation,#labelling,#prosody,#alignment,#AMPER## # SEA_AP:!UNA!HERRAMIENTA!DE!SEGMENTACIÓN!Y!ETIQUETADO!PARA!EL!ANÁLISIS!PROSÓDICO! Resumen! En# este# artículo# se# presenta# una# herramienta# que# realiza# la# segmentación# y# el# etiquetado# de# cadenas# sonoras# en# unidades# de# fono,# sílaba# y/o# palabra# partiendo# de# una# señal# sonora# y# de# su# P.!LÓPEZ!OTERO!et!al.! 224# correspondiente#transcripción#ortográfica.#Además,#integra#scripts#de#análisis#acústico#que#se#ejecutan# sobre#el#programa#Praat#con#el#fin#de#reducir#el#tiempo#invertido#en#las#tareas#de#análisis,#corrección,# suavizado# y# generación# de# gráficos# de# la# curva# melódica.# La# herramienta# está# implementada#para# gallego,#español#y#portugués#de#Brasil.#Nuestro#objetivo#es#contribuir#con#esta#aplicación#a#automatizar# algunas#de#las#labores#de#segmentación,#etiquetado#y#análisis#prosódico,#pues#constituyen#tareas#que# requieren#una#gran#inversión#de#tiempo#y#de#recursos#humanos.# # Palabras!clave! segmentación,#etiquetado,#prosodia,#alineamiento,#AMPER# # # 1.!Introduction! # The#study#of#prosody,#and#intonation#in#particular,#focuses#on#the#description#of# the#evolution#of#the#fundamental#frequency,#the#duration#and#the#intensity#linked#to# the#sound#chains#which#convey#linguistic#meaning.#One#of#the#fields#of#study#of#the# prosody# is# the# analysis# of# linguistic# variation# (geographical,# social# or# contextual,# mainly).# In# recent# years,# the# study# of# the# geoprosodic# variation# has# received# considerable#attention#from#researchers,#so#that#today#we#have#a#significant#number#of# prosodic#atlas.#In#the#Romance#field,#we#should#highlight#the#works#carried#out#in#the# framework#of#two#projects:#the#Atlas!Multimedia!de!la!Prosodia!del!Espacio!Románico#-# AMPER# (http://dialecto.u-grenoble3.fr/AMPER/amper.htm)# and# the# Interactive! Atlas! of! Romance! Intonation# -# IARI# (http://prosodia.upf.edu/iari/index.html).# In# both# projects#a#great#number#of#European#and#American#universities#and#research#groups# collaborate.# In# both# cases,# there# are# numerous# results# which# are# available# on# the# different#websites#(general#and#specific#for#each#group)#as#well#as#in#publications#of# various#kinds#(Mairano#2011,#which#also#includes#a#large#number#of#references;#Frota#&# Prieto#2015).# We#are#particularly#interested#in#the#case#of#AMPER#(Contini#et#al.#2002),#as#it# collects# data# from# all# the# Iberian# varieties# (Galician,# Asturian-Leonese,# Portuguese,# Castilian#Spanish,#Catalan,#etc.).#Furthermore,# in# the# case# of# this# atlas,#it#provides# a# detailed# phonetic# description# of# the# intonation# contours# of# declarative# and# interrogative#modalities,#which#has#enabled#not#only#an#important#database#based#on# Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 225# data#mapping,#but#it#has#also#facilitated#the#development#of#other#methodologies#of# geolinguistic#study,#such#as#the#dialectometry#(Fernández#Planas#et#al.#2011#and#2015,# Martínez# Calvo# &# Fernández# Rei# 2015,# Moutinho# et# al.# 2011).# Behind# all# these# outcomes,#there#is#intense#work#comprising#recording,#codification#and#analysis#at#a# high#cost,#both#in#terms#of#human#resources#and#time.#The#tasks#of#labelling#audio#files# (from#which#the#prosodic#data#which#allows#us#to#perform,#amongst#other#tasks,#data# analysis# or# mapping,# is# obtained)# are# only# partially# automatized,# making# it# a# very# strenuous# process,# both# in# developing# routines# for# Matlab,# by# Antonio# Romano# (http://www.lfsag.unito.it/amper/fox.html)# or# by# the# group# AMPER# Asturias# (http://www.unioviedo.es/labofone/),# as# in# the# preparation# of# scripts# in# Praat# (Boersma# &# Weenik# 2013),# in# which# important# contributions# have# been# made# by# Albert# Rilliard# (https://perso.limsi.fr/rilliard/InterfaceAMPER.html)# and# the# AMPER# group# of# the# Universitat# de# Barcelona# (Martínez# Celdrán# &# Fernández# Planas# 20032015).# This# project# began# circa# 2000,# so# that# during# almost# fifteen# years,# prosodic# dialectal# data# has# been# gathered# and# analysed.# Each# group# has# tried# to# solve# the# problems#that#have#arisen#and#there#have#been#many#advances#from#which#all#of#the# groups# have# benefited.# Nevertheless,# there# has# also# been# some# fragmentation# and# dispersal# of# the# resources# and# tools# developed.# The# need# therefore# arises# to# unify# these#resources#in#packages#by#taking#the#advantages#jointly#offered.## With#regard#to# the#tools# for#the#alignment#of#audio#and#text#with#subsequent# labelling#of#results,#various#studies#have#appeared,#such#as#EasyAlign#(Goldman#2011),# aimed#at#the#alignment#of#audio#and#text#and#with#support#for#Spanish,#English,#French,# Brazilian#Portuguese#and#Taiwanese#and#which,#in#addition,#is#run#using#Praat.#There#is# also# SegProso# (Garrido# 2013),# which# is# a# Praat-based# tool# for# the# automatic# segmentation#of#speech#corpora#into#prosodic#units#for#Spanish#and#Catalan#that#need# initial#TextGrid#files#which#must#contain#at#least#two#tiers,#the#first#of#which#with#the# word# segmentation# and# a# second# tier# including# a# phonetic# transcription# in# SAMPA# format.# This# tool# provides# a# rule-based# segmentation# and# linguistic# knowledge.# Another#useful#tool#is#SPPAS#(Bigi#&#Hirst#2012)#to#automatically#produce#annotations# which#include#utterance,#word,#syllable#and#phoneme#segmentation#with#support#for# P.!LÓPEZ!OTERO!et!al.! 226# French,# English,# Italian# and# Chinese;# output# can# be# used# later# by# other# tools# for# prosodic#analysis.#In#addition,#the#AuToBI#Rosenberg#(2009,#2010)#tool#for#automatic# generation#of#hypothesized#ToBI#labels,#which#has#been#applied#to#Portuguese#(Moniz# et#al.#2014).# However,#for#many#minority#languages#with#a#deficit#of#linguistic#resources,#such# as# Galician,# a# tool# capable# of# automating# these# tasks# has# not# yet# been# developed,# sometimes#due#to#the#lack#of#acoustic#models#or#to#the#lack#of#other#resources#such#as# grapheme-phoneme#converters.#Currently,#specifically#for#Galician,#there#was#no#tools# at#user#level#that#would#allow#to#use#the#acoustic#models#obtained#by#the#Multimedia# Technology# Group# of# the# Universidade# de# Vigo# for# this# language,# combined# with# a# grapheme-phoneme#converter#implemented#by#the#same#group#mentioned#before#in# collaboration#with#linguists#on#the#project#Síntese!de!voz!of#the#Ramón#Piñeiro#centre# and#with#one#recogniser#for#performing#an#alignment.#In#this#case,#the#recogniser#used# is#provided#by#the#HTK#package.#For#this#reason,#the#system#board#combines#different# resources#in#a#single#tool#usable#by#any#user#without#experience#in#speech#processing;# the# diverse# resources# were# mainly# tailored# to# the# needs# of# the# AMPER# project,# because#this#is#a#large#big#database#in#different#romance#languages#in#which#the#data# concerned#follows#a#specific#structure.#The#main#novelty#provided#in#this#work#is#the# development#of#an#automatic#tool#which#comprises#many#resources.#Moreover,#this# programme# is# the# first# of# its# kind# which# includes# segmentation# and# labelling# for# Galician,# providing# its# users# with# an# improved# tool# in# terms# of# time# spent# on# these# tasks.## From# this# point# of# view,# there# are# three# important# aspects# which# should# be# strengthened:# 1.#Automatic#segmentation#and#vowel#labelling,#which#will#be#the#basis#on#which# the#prosodic#analysis#is#sustained.# 2.#The#integration#of#existing#tools.# 3.#The#tool#is# currently# implemented# to# be# employed#primarily#by# researchers# working#within#the#framework#of#AMPER#methodology.# In#this#way,#in#this#paper#we#present#a#package#that#enables#the#alignment#of# audio# and# text# devoid# of# timestamps,# in# addition# to# subsequent# vowel# labelling.# Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 227# Furthermore,# support# for# other# languages# with# available# voice# recordings# of# appropriate#quality#or#acoustic#models#has#also#been#implemented:#Galician,#Spanish# and#Brazilian#Portuguese.#The#addition#of#more#languages#is#expected#in#the#future,# particularly#the#remaining#Romance#varieties,#although#it#would#be#necessary#to#have# acoustic# models# or# quality# speech# databases# and# grapheme-phoneme# converters# in# the# language# intended# to# be# incorporated.# This# integration# of# a# larger# number# of# linguistic#varieties#would#greatly#increase#the#AMPER#database#and,#therefore,#allow#a# much#more#complete#dialectal#map#of#Romance#intonation.## On#the#other#hand,#a#second#objective#achieved#with#this#work#has#been#to#seek# the# integration# of# all# the# tools# most# employed# by# the# groups# participating# in# the# AMPER# project# for# the# extraction# of# F0# data,# duration# and# energy,# generation# of# graphics,# etc.,# that# provide# a# prosodic# analysis# of# the# recordings.# In# the# future,# the# integration# of# R# scripts# which# provide# dialectrometric# analysis# (Martínez# Calvo# &# Fernández#Rei#2015)#is#also#intended.# With# this# tool,# named# SEA# AP,# the# automation# of# certain# tasks# that# were# previously# performed# manually,# especially# in# data# acquisition# (measuring,# labelling,# transferring#to#an#Excel#sheet,#etc.),#is#enabled.#Thus#we#are#witnessing#an#extraction# and#storage#of#data#performed#automatically#and#which#will#be#later#used#for#study# and#analysis.#The#tool#is#aimed#at#researchers#who#seek#to#focus#their#efforts#on#the# analysis# and# interpretation# of# data,# rather# than# on# segmentation,# labelling# and# obtainment#of#prosodic#data.## The#tool#is#freely#available#for#use#by#the#entire#research#community.#To#access# the# repository,# the# Multimedia# Technology# Group# of# the# Universidade# de# Vigo,# the# Instituto#da#Lingua#Galega#of#the#Universidade#de#Santiago#de#Compostela,#or#any#of# the#authors#of#this#paper#should#be#contacted.# This#paper#describes#the#details#of#the#development#and#use#of#the#tool.#First,#it# provides#a#brief#description#of#the#tool#explaining#its#characteristics#and#some#aspects# with#regard#to#the#internal#operation.#Then,#there#is#an#explanation#of#the#module#in# charge# of# the# segmentation# and# labelling# of# the# input# material.# In# the# following# section,# the# scripts# used# for# the# prosodic# analysis# integrated# there# are# described.# P.!LÓPEZ!OTERO!et!al.! 228# Finally,#the#conclusions#obtained#from#this#work#and#the#resulting#future#lines#are#set# out.# ! 2.!Tool!features! # SEA_AP# (Segmentador! e! Etiquetador! Automático! para! Análise! Prosódica,! Automatic!Segmentation!and!Labelling!for!Prosodic!Analysis)#is#a#desktop#application,# compatible#with#the#Windows#operating#system,#allowing#the#alignment#of#audio#files# with# their# respective# transcription,# as# well# as# their# prosodic# analysis.# The# alignment# results#produces#segmentation#with#temporal#labelling#with#regard#to#words,#syllables# and/or#phonemes.#In# the# main#window#of# the# application,# the#type# of# units#for#the# segmentation,#as#well#as#the#language#to#analyse#the#data,#can#be#selected.## After#this#first#phase#of#alignment,#the#application#generates#the#data#needed#for# a# prosodic# analysis# using# the# Praat# software;# it# will# therefore# be# run# along# with# integrated# scripts# performing# interactions# with# the# user# so# that# he/she# can# modify# certain#parameters#used#in#the#analysis,#if#desired,#as#well#as#the#visualization#of#the# results#obtained#from#the#analysis#executed.## The#data#that#will#be#provided#for#the#tool#is#the#directory#where#the#audio#files# are,#the#directory#with#the#texts#to#be#aligned,#the#directory#where#we#would#like#the# TextGrid# for# Praat# to# be# generated# and,# finally,# the# directory# for# storing# the# results# from#the#prosodic#analysis.#All#this#information#is#required#for#the#proper#operation#of# the#tool.# Once#all#this#input#info#has#been#entered,#the#segmentation#and#labelling#tasks# will#start#when#the#alignment#button#is#clicked.#In#order#to#carry#them#out,#the#first# step#performed#is#an#automatic#phonetic#transcription#of#a#given#text#in#.txt#format,# transforming# the# graphemes# in# phonemes.# Subsequently,# a# forced# recognition# assigning# the# aforementioned# units# to# audio# segment# by# means# of# the# recognition# engine# HTK# (Hidden# Markov# Model# Toolkit)# (Young# et# al.# 1995)# will# be# performed.# Finally,# the# scripts# for# the# prosodic# study# will# be# executed# based# on# the# previous# segmentation#and#labelling.## # Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 229# # # 3.#Elements!of!the!alignment!block! # This#section#will#deal#with#a#brief#description#of#the#different#elements#that#make# the# alignment# of# the# sound# waveform# and# the# orthographic# transcription# provided# possible.## Firstly,# for# the# automatic# phonetic# transcription# for# Galician# and# Spanish# languages,#the#Cotovia#package#has#been#used#(González#2004,#Rodríguez#Banga#et#al.# 2012),# which# was# developed# by# the# centre# Ramón# Piñeiro# in# collaboration# with# the# Universidade# de# Vigo# and# the# Universidade# de# Santiago# de# Compostela,# groups# to# which#the#authors#belong.#The#system#used#for#Portuguese#has#been#obtained#from#the# FalaBrasil#project#(Nelson#et#al.#2010),#developed#by#the#Universidade#Federal#do#Pará.## Secondly,#the#system#for#labelling#phonemes#is#implemented#by#using#the#HTK# tool,# consisting# of# a# series# of# modules# that# perform# the# necessary# tasks# for# the# implementation#of#recognisers#based#on#HMM#(Torre#&#Hernández#2002).# # 3.1!Automatic!phonetic!transcription!! # As# mentioned# above,# the# first# step# is# to# obtain# an# automatic# phonetic# transcription#(González#et#al.#2008)#of#the#text#to#be#aligned#with#the#sound#waveform.# For#this#task,#a#number#of#steps#are#performed:## a)#Linguistic#Preprocessing:#the#texts#usually#contain#words#such#as#acronyms#or# numbers#that#needs#to#be#normalised#so#that#the#conversion#is#correctly#completed.# For#this,#the#following#steps#are#carried#out:## -#An#expansion#of#the#abbreviations#and#acronyms#contained#in#the#sequence.# -# The# replacement# of# numbers,# both# Arabic# and# Roman,# using# certain# reading# rules,#as#well#as#the#transcription#of#time#expressions.# -#The#transcription#of#foreign#words#by#adjusting#the#sounds#used#in#the#language# studied,#since#as#a#general#rule#grapheme-sound#correspondence#is#particular#to#each# language.# P.!LÓPEZ!OTERO!et!al.! 230# b)#A#morphological#analysis:#this#involves#the#classification#of#words#according#to# their#grammatical#category,#gender#and#number#as#described#in#Seijo#et#al.#(2004).#The# steps#to#be#taken#to#perform#this#classification#are#the#following:## -# The# labelling# of# words# belonging# to# grammatical# groups# with# low# variability,# such#as#prepositions,#adverbs,#conjunctions#or#determinants.## -#An#analysis#of#the#roots#and#morphemes#of#words#with#greater#variability,#such# as#names,#adjectives#or#verbs,#by#assigning#them#all#possible#categories#from#amongst# which#the#definitive#one#will#be#chosen#afterwards.#This#disambiguation#is#carried#out# by#means#of#a#contextual#analysis#which#takes#into#account#the#categories#that#were# already#assigned#to#the#preceding#and#following#words.#In#addition,#a#statistical#model# which#takes#into#consideration#the#most#probable#category#is#also#applied.## -#The#study#of#word#groups#such#as#adverbial#or#conjunctive#phrases#and#verbal# periphrasis.## After# categories# are# assigned,# words# are# divided# into# syllables# by# using# an# algorithm,#implemented#by#the#Cotovia#package#(González#2004,#Rodríguez#Banga#et# al.#2012),#which#detects#the#syllable#nucleus#(which#is#always#a#vowel#in#the#case#of#the# languages#supported#by#the#application)#and#which#takes#into#account#the#surrounding# consonants#in#order#to#determine#the#syllable#boundaries.#It#is#also#at#this#point#when# stress#is#assigned.#The#syllabication#algorithm#takes#into#consideration#diphthongs#and# triphthongs#and#allows#the#detection#of#less#frequent#phoneme#sequences#that#appear# in#words#from#Latin,#Greek#or#other#foreign#languages.# c)#A#syntactic#analysis:!this#step#is#focused#on#the#analysis#of#the#sentences#taking# into#account#the#existing#prepositions#and#the#way#in#which#the#words#and#syntagms# relate#to#each#other.#The#position#of#punctuation#signs#and#words#such#as#conjunctions,# conjunctive#phrases#or#relative#pronouns#is#also#taken#into#account.# It# also# takes# into# account# the# position# of# punctuation# and# words# such# as# conjunctions,#conjunctive#phrases#or#relative#pronouns.# In#addition,#a#re-accentuation#of#words#is#accomplished#by#removing#the#stress#of# function#words#and#by#inserting#pauses.#The#latter#are#inserted#through#to#punctuation# signs#and#in#sequences#exceeding#a#certain#length.## Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 231# d)# A# phonetic# transcription:!In# this# phase# the# grapheme# string# becomes# a# sequence#of#phonetic#characters#representing#the#allophones.## # 3.2!Automatic!Labelling!of!Phonemes! # Once#the#allophonic#representation#of#the#text#is#available,#the#next#step#will#be# to#obtain#their#occurrence#temporal#instants,#for#this#purpose#an#automatic#labelling#is# performed.# A#block#diagram#of#the#system#used#is#shown#below:# # # Figure#1.#Diagram#of#the#alignment#module# # As# shown,# the# system# mainly# consists# of# four# modules# which# are# described# below:## a)#Acoustic#features#module:#the#system#takes#an#audio#signal#as#input,#and#this# module#is#responsible#for#extracting#the#features#which#intend#to#model#the#behaviour# that#the#human#ear#would#display#in#the#presence#of#a#similar#stimulus.#In#this#case,#the# features#used#are#the#Mel-Frequency!Cepstral!Coefficients,#MFCCs.-These#coefficients# are#suitable#given#that#they#adequately#simulate#the#human#auditory#perception#and,# furthermore,#they#are#robust#against#background#noise.#- b)#Acoustic#modelling#module:-this#module#is#the#responsible#for#relating#certain# sound# waveforms# with# the# corresponding# allophone,# with# each# allophone# therefore# represented# as# a# sound.# However,# a# phoneme# has# many# variations# given# that# one# person# does# not# always# pronounce# a# word# in# the# same# way,# different# people# have# P.!LÓPEZ!OTERO!et!al.! 238# # Figure#4.#Data#entry#form#for#prosodic#analysis# # In#this#form,#as#mentioned#above,#the#system#automatically#fills#the#fields#of#the# directories#where#the#audio#files#and#the#associated#TextGrid#are#located,#with#the#data# entered# by# the# user# in# the# first# form.# Moreover,# the# user# must# select# the# types# of# analysis#desired,#either#checking#or#unchecking#the#boxes#provided#for#that#purpose.## Once#the#user#clicks#OK#on#the#form,#the#scripts#selected#will#begin#to#run.#In#any# case,# the# user# is# informed# about# the# step# to# be# executed# and# provided# with# the# possibility#to#skip#it,#if#desired,#running#then#the#following#step#of#the#list.#(Figure#5).# # # Figure#5.#Information#window#displayed#to#the#user#before#each#step# # Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 239# For#the#review#task,#the#TextGrid#generated#in#the#first#step#is#shown,#which#only# contain#the#phonetic#segmentation.#Thus,#at#this#stage#of#manual#review,#the#user#only# has#to#correct#that#layer,#from#which,#once#corrected,#the#system#will#generate#the#rest# of#the#lines.#In#the#event#that#the#user#does#not#want#to#launch#this#step,#the#system# will# generate# the# remaining# lines# with# the# available# segmentation# in# syllables# and# words.# In#the#task#of#“Graph#Generation”,#the#form#is#shown#in#Figure#6,#where#the#user# can# select# the# tiers# that# will# appear# on# the# graph# generated# from# the# TextGrid,# avoiding#the#display#of#those#not#considered#interesting#for#their#purposes.## # # Figure#6.#Window#for#the#selection#of#tiers#to#be#shown#in#the#graph#within#the#TextGrid# # The#remaining#steps#perform#the#tasks#described#in#the#scripts#of#the#previous# section.## # # 6.!Conclusions!! # The#work#presented#in#this#project#has#required#the#collaboration#of#numerous# research#sources#from#different#areas#carried#out#at#different#universities,#seeking#to# attain#the#ultimate#objective#of#integrating#of#all#the#improvements#provided#by#each# of#them#in#a#single#robust#and#easy#to#use#tool.# P.!LÓPEZ!OTERO!et!al.! 240# Therefore,#based#on#the#needs#expressed#by#the#specialists#on#prosodic#analysis,# a#search#has#been#performed#of#the#most#useful#tools#which#are#the#result#of#several# research# activities,# and# that,# grouped# together,# constitute# a# very# beneficial# contribution#in#order#to#automate#certain#works#that#were#otherwise#very#laborious.## Finally,# as# a# result# of# the# cooperation# between# the# engineering# and# linguistic# areas,#and#thanks#also#to#the#improvements#made,#we#have#been#able#to#automate#the# time-consuming# manual# segmentation# and# labelling# by# means# of# an# easy-to-use# desktop# application,# tailored# to# the# needs# of# the# target# users,# which# allows# a# considerable#saving#on#the#time#spent.#Furthermore,#it#brings#together#segmentation,# labelling#and#prosodic#analysis#in#a#single#tool.## # # 7.!Future!work! # We# are# dealing# with# a# complete# and# ready# to# use# tool.# However,# as# a# continuation#of#this#project,#we#may#propose#to#undertake#a#case#study#to#evaluate#the# performance#of#the#alignment#by#means#of#a#comparison#between#a#manually#labelled# database#and#a#labelling#performed#with#SEA_AP.## Given#that#the#performance#of#the#alignment#largely#relies#on#the#quality#of#the# forced# recognition# and,# therefore,# on# the# acoustic# models,# it# is# also# possible# to# improve#those#models#by#adapting#them#to#the#material#aligned#using#this#tool.#Thus,# the#quality#of#the#next#alignment#and#the#existing#repositories#for#a#particular#language# in#the#minority#languages#would#be#improved.# In# addition,# this# tool# could# be# useful# in# less# controlled# scenarios# than# those# originally#conceived,#since#it#could#be#applied#in#spoken#language#corpus#of#different# levels#and#registers:#rural#language,#urban#language,#formal#language,#etc.#(Escourido# et# al.# 2008).# An# example# of# this# type# of# corpus# would# be# CORILGA:# Corpus! Oral! Informatizado!de!la!Lengua!Gallega#(García-Mateo#et#al.#2014).#This#way,#a#prosodic# labelling#of#the#material#collected#in#this#corpus#could#be#provided,#which#would#be#a# significant# advance# as# it# would# allow# us# to# address# the# study# of# intonation# in# Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 241# spontaneous#interactions,#one#of#the#major#challenges#in#current#research#in#the#field# of#prosodic#studies.## Another# potential# line# of# research# is# the# application# of# a# methodology# of# dialectometric#analysis#to#the#prosodic#data#generated#by#our#tool.#To#the#extent#that# SEA_AP# allows# the# labelling# and# analysis# of# large# quantities# of# sequences,# the# integration#of#scripts#for#the#dialectometric#analysis#of#the#results#(Martínez#Calvo#&# Fernández#Rei#2015)#could#be#proposed.## Finally,#it#is#worth#noting#the#possibility#of#expanding#this#to#other#languages#with# sufficient#data#available#in#order#to#train#the#acoustic#model#and#a#suitable#graphemephoneme#converter,#and#to#also#create#a#multilingual#user#interface.# # # 8.!Acknowledgements! # This# work# would# have# not# been# possible# without# the# help# of# the# Spanish# Government# (Project# ‘SpeechTech4All’# TEC2012-38939-C03-01),# the# European# Regional#Development#Fund#(ERDF),#the#Government#of#the#Autonomous#Community# of# Galicia# (GRC2014/024,# “Consolidación# de# Unidades# de# Investigación:# Proyecto# AtlantTIC”#CN2012/160)#and#the#“Red#de#Investigación#TecAnDAli”#from#the#Council#of# Culture,#Education#and#University#Planning,#Xunta#de#Galicia.## We# would# also# like# to# thank# the# anonymous# reviewers# for# their# suggestions# which#contributed#to#improving#the#article.# # # References! # BARBOSA,#P.A.#(2006)#Incursões!em!torno!do!ritmo!da!fala,#Campinas:#Pontes.# BIGI,#B.#&#D.#HIRST#(2012)#“Speech#Phonetization#Alignment#and#syllabification#(SPPAS):#a#tool# for#the#automatic#analysis#of#speech#prosody”,#Proceedings!of#Speech!Prosody,#Shanghai,# 1-4.#<https://hal.archives-ouvertes.fr/hal-00983699># P.!LÓPEZ!OTERO!et!al.! 242# BOERSMA,# P.# &# D.# WEENIK#(2013)# Praat:! doing! phonetics! by! computer#[Computer# program],# version#5.4.05#[http://www.praat.org/].# CONTINI,#M.,#J.P.#LAI#&#A.#ROMANO#(2002)#“La#géolinguistique#à#Grenoble:#de#l’AliR#à#l’AMPER”,#in# M.R.#Simoni-Aurembou#(ed.),#Nouveaux!regards!sur!la!variation!diatopique,!Revue!belge! de!Philologie!e!d’Histoire,#80,#931-941.# DOCIO-FERNÁNDEZ,# L.,# A.# CARDENAL-LÓPEZ,# C.# GARCÍA-MATEO#(2006)# “TC-STAR# 2006# Automatic# Speech#Recognition#Evaluation:#The#UVIGO#System”,#in#TC-STAR!Workshop!on!Speech-toSpeech!Translation,#Barcelona,#145-150.## #<https://www.researchgate.net/publication/255661612_TCSTAR_2006_Automatic_Speech_Recognition_Evaluation_The_UVIGO_System># ELVIRA-GARCÍA,#W.#(2014a)#Prosodic-data-extraction!v2.1#[Praat#script,#distributed#under#GNU# General#Public#License]#<http://stel.ub.edu/labfon/en/praat-scripts># ELVIRA-GARCÍA,# W.# (2014b)# Blank_TextGrid_creation#[Praat# script,# distributed# under# GNU# General#Public#License]#<http://stel.ub.edu/labfon/en/praat-scripts># ELVIRA-GARCÍA,# W.# (2014c)# Remove_tiers#[Praat# script,# distributed# under# GNU# General# Public# License]#<http://stel.ub.edu/labfon/en/praat-scripts># ELVIRA-GARCÍA,#W.#(2014d)#Created_TextGrid_modification#[Praat#script,#distributed#under#GNU# General#Public#License]#<http://stel.ub.edu/labfon/en/praat-scripts># ELVIRA-GARCÍA,#W.#&#P.#ROSEANO#(2014)#Create!pictures!with!tiers!v.4.1.#[Praat#script,#distributed# under#GNU#General#Public#License]#<http://stel.ub.edu/labfon/en/praat-scripts># ESCOURIDO,#A.,#E.#FERNÁNDEZ#REI,#M.#GONZÁLEZ#&#X.#L.#REGUEIRA#(2008)#“A#dimensión#prosódica#da# oralidade.# Achega# dende# AMPER”,# in# E.# Fernández# Rei# &# X.L.# Regueira,# Perspectivas! sobre! a! oralidade,# Santiago# de# Compostela:# Instituto# da# Lingua# Galega# /# Consello# da# Cultura#Galega,#75-93.# FERNÁNDEZ#PLANAS,#A.M.,#P.#ROSEANO,#E.#MARTÍNEZ-CELDRÁN#&#L.#ROMERA#(2011)#“Aproximación#al# análisis# dialectométrico# de# la# entonación# en# algunos# puntos# del# dominio# lingüístico# catalán”,#Estudios!de!Fonética!Experimental,#XX,#141-178.# FERNÁNDEZ#PLANAS,#A.M.,#J.#DORTA,#P.#ROSEANO,#Ch.#DÍAZ,#W.#ELVIRA-GARCÍA#&#E.#MARTÍNEZ-CELDRÁN# (2015)# “Distancia# y# proximidad# prosódica# entre# algunas# variedades# del# español:# un# estudio# dialectométrico# a# partir# de# datos# acústicos”,# Revista! de! Lingüística! Teórica! y! Aplicada,#53#(2),#13-45.# FROTA,#S.#&#P.#PRIETO#(eds.)#(2015)#Intonational!Variation!in!Romance,#Oxford:#Oxford#University# Press.# Dialectologia.!Special-issue,-VI-(2016),!223-244.!! ISSN:!2013-2247! 243# GARCIA-MATEO,# C.,# J.# DIEGUEZ-TIRADO,# A.# CARDENAL-LOPEZ,# &# L.# DOCIO-FERNANDEZ#(2004)# “Transcrigal:# A# bilingual# system# for# automatic# indexing# of# broadcast# news”,# in# Proceedings!Int.!Conf.!on!Language!Resources!and!Evaluation,#volume#6,#Lisbon:#ELRA,# European# Language# Resources# Association,# 2061-2064.#<http://www.lrecconf.org/proceedings/lrec2004/pdf/382.pdf># GARCÍA-MATEO,#C.,#A.#CARDENAL,#X.L.#REGUEIRA#FERNÁNDEZ,#E.#FERNÁNDEZ#REI,#M.#MARTÍNEZ,#R.#SEARA,# R.#VARELA#&#N.#BASANTA#LLANES#(2014)#“CORILGA:#a#Galician#Multilevel#Annotated#Speech# Corpus# for# Linguistic# Analysis”,# in# N.# Calzolari,# K.# Choukri,# T.# Declerck,# H.# Loftsson,# B.# Maegaard,#J.#Mariani,#A.#Moreno,#J.#Odijk#&#S.#Piperidis#(eds.),#Proceedings!of!the!Ninth! International! Conference! on! Language! Resources! and! Evaluation! (LREC'14),# Reykjavik:# ELRA#<http://www.lrec-conf.org/proceedings/lrec2014/index.html># GARRIDO,# J.# M.# (2013)# “SegProso:# A# Praat-Based# Tool# for# the# Automatic# Detection# and# Annotation# of# Prosodic# Boundaries”,# Proceedings! of#TRASP,# 74-77.# <http://www.lplaix.fr/~trasp/Proceedings/19864-trasp2013.pdf># GOLDMAN,#J.# P.# (2011)# “EasyAlign:# an# automatic# phonetic# alignment# tool# under# Praat”,# Proceedings!of!InterSpeech,#Firenze,#Italy,#3233-3236.## #<http://latlcui.unige.ch/phonetique/easyalign/easyalign_unpublished.pdf># GONZÁLEZ,#M.#(2004)#“A#síntese#de#voz#en#lingua#galega:#O#proxecto#Cotovía”,#Revista!galega!do! ensino,#44,#199-215.# GONZÁLEZ# GONZÁLEZ,# M.,# E.# RODRÍGUEZ# BANGA,# F.# CAMPILLO# DÍAZ,# F.# MÉNDEZ# PAZÓ,# L.# RODRÍGUEZ# LIÑARES#&# G.# IGLESIAS# IGLESIAS#(2008)# “Specific# features# of# the# Galician# language# and# implications#for#speech#technology#development”,#Speech!Communication,#50,#874-887.# MAIRANO,#P.#(ed.)#(2011)#Intonations!Romanes,#Géolinguistique,#hors-série#4.# MARTÍNEZ# CALVO,# A.# &# E.# FERNÁNDEZ# REI#(2015)# “Unha# ferramenta# informática# para# a# análise# dialectométrica# da# prosodia”,# Estudios! de! Fonética! Experimental,# XXIV,# 289-303# <http://stel.ub.edu/labfon/sites/default/files/9_MARTINEZ.pdf># MARTÍNEZ#CELDRÁN,#E.#&#A.#M.#FERNÁNDEZ#PLANAS,#(coords.)#(2003-2015)#Atlas!Multimèdia!de!la! Prosòdia!de!l’Espai!Romànic.# <http://stel.ub.edu/labfon/amper/cast/index_ampercat.html># MONIZ,#H.,#A.I.#MATA,#J.#HIRSCHBERG,#F.#BATISTA,#A.#ROSENBERG#&#I.#TRANCOSO#(2014)#“Extending# AuToBI#to#prominence#detection#in#European#Portuguese”,#In#N.#Campbell,#D.#Gibbon#&# D.# Hirst# (eds.),# Proceedings! of! Speech! Prosody,# Dublin,#Trinity# College,#280-284# <http://www.speechprosody2014.org/># P.!LÓPEZ!OTERO!et!al.! 244# MOUTINHO,# L.C.,# R.L.# COIMBRA,# A.# RILLIARD#&# A.# ROMANO#(2011)# “Mesure# de# la# variation# prosodique#diatopique#en#portugais#européen”,#Estudios!de!Fonética!Experimental,#20,# 33-55.# NELSON#NETO,#P.#SILVA,#A.#KLAUTAU#&#I.#TRANCOSO#(2010)#“Free#tools#and#resources#for#Brazilian# Portuguese# speech# recognition”,# Journal! of! the! Brazilian! Computer! Society# <http://link.springer.com/article/10.1007%2Fs13173-010-0023-1># RABINER,#L.R.#(1989)#“A#tutorial#on#hidden#Markov#models#and#selected#applications#in#speech# recognition”,#Proceedings!of!the!IEEE,#Vol.#77,#No.#2,#257-286.# <http://www.cs.ubc.ca/~murphyk/Bayes/rabiner.pdf># RILLIARD,# A.# (2013)# “Metodoloxía# cuantitativa# para# a# medida# das# distancias# prosódicas”,# Xornadas! de! Dialectoloxía! Perceptiva,# Santiago# de# Compostela# <http://ilg.usc.es/tecandali/Descargas/AlbertRilliard.pdf># RODRÍGUEZ#BANGA,#E,#C.#GARCÍA#MATEO,#F.#J.#MÉNDEZ#PAZÓ,#M.#GONZÁLEZ#&#C.#MAGARIÑOS#IGLESIAS# (2012)# “Cotovía:# an# open# source# TTS# for# Galician# and# Spanish”,# VII# Jornadas# en# Tecnología# del# Habla# and# III# Iberian# SLTech# Workshop,# IberSPEECH! 2012,# Madrid# <http://iberspeech2012.ii.uam.es/IberSPEECH2012_OnlineProceedings.pdf># ROSENBERG,# A.# (2009)# “Automatic#detection# and# classification# of# prosodic# events”,# Columbia# University,# Ph.D.# Thesis.# <http://www1.cs.columbia.edu/~amaxwell/amaxwell-thesisfinal.pdf># ROSENBERG,#A.#(2010)#“AuToBI-A#tool#for#automatic#ToBI#annotation”,#in#INTERSPEECH,#146-149.# <http://eniac.cs.qc.cuny.edu/andrew/papers/autobi-is10.pdf># SEIJO# PEREIRO,# L.,# A.#MARTÍNEZ#ÍNSUA,# F.# MÉNDEZ# PAZÓ,# F.# CAMPILLO# DÍAZ#&# E.# RODRÍGUEZ# BANGA# (2004)#“A#Galician#Textual#Corpus#for#Morphosyntactic#Tagging#with#Application#to#Textto-Speech# Synthesis”,# in# Proceeding! of! LREC#2004,# Lisboa,#vol.# 5,# 1759-1762.# <http://www.lrec-conf.org/proceedings/lrec2004/pdf/111.pdf># TORRE# TOLEDANO,# D.# &# L.# HERNÁNDEZ# GÓMEZ#(2002)# “Hmms# for# automatic# phonetic# segmentation”,# in# Proc.! of! LREC#2002,# Las# Palmas# de# Gran# Canaria# [http://www.lrecconf.org/proceedings/lrec2002/].# YOUNG,#S.,#G.#EVERMANN,#M.#GALES,#T.#HAIN,#D.#KERSHAW,#G.#MOORE,#J.#ODELL,#D.#OLLASON,#D.#POVEY,# V.# VALTCHEV#&# P.# WOODLAND#(1995)# HTK! Book,# University# of# Cambridge# <http://htk.eng.cam.ac.uk/docs/docs.shtml># #