voxbi 0.1.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA1:
+  metadata.gz: 6ef2064ad85e3c2ed71c515df6c0d6882c20830a
+  data.tar.gz: 409ca543dfb46f1f4cdd4c3acfbf0cfcb9133e6f
+SHA512:
+  metadata.gz: 41f37cbdbdbe9e49d81cf4e2054e789e813141bb0d80423e68bbd1aea3c973d73a3e24b8cec22ed89b830676d241df4c6a64f53e328e504e971381545ff9b3a4
+  data.tar.gz: c3adf1948be1c5a3ad16fd21473b820e412844a8564fa566e6ee6906e0e1650970e49e96a3f336c19f016e0a09237b92cf9250d91d44b39661197fc4958189fd

data/bin/voxbi

@@ -0,0 +1,46 @@
+#!/usr/bin/env ruby
+
+ROOT = File.expand_path("../..", __FILE__)
+
+def parseCSV(path)
+	Hash[File.open("#{ROOT}/data/#{path}.csv").read.split("\n").map {|ligne| ligne.split("#")}]
+end
+
+def exceptions
+	@exceptions ||= parseCSV "exceptions"
+end
+
+def conversion
+	@conversion ||= parseCSV "conversion"
+end
+
+def apimatch(texte)
+	texte.gsub(/-'",:;\.\(\)/, "").split.map do |mot|
+		exceptions[mot] || "".tap do |result|
+			conversion.select { |regle| mot =~ /#{regle}/ }.first.tap do |regle, api|
+				mot.sub! /#{regle}/, ""
+				result << api.to_s
+			end until mot.empty?
+		end
+	end
+end
+
+def voxbi(texte)
+	paires_dispo = File.open("#{ROOT}/data/paires_disponibles.csv").read.split("\n")
+	api = apimatch(texte).join "_"
+	puts api.inspect
+	fichiers = []
+	while api.length !=0
+		paires_dispo.each do |paires|
+			if api.match(/^#{paires}/)
+				fichiers << "#{ROOT}/data/paires/#{paires}.ogg"
+				api = api.sub(/^#{paires}/,"")
+				break
+			end
+		end
+	end
+	`sox #{fichiers.join(" ")} #{ROOT}/data/paires.wav`
+	`aplay  #{ROOT}/data/paires.wav`
+end
+
+voxbi(ARGV[0])

data/data/conversion.csv

@@ -0,0 +1,152 @@
+^ch$#k
+^uë#y
+^oë#oɛ
+^s(?=['ç'])#s
+^ueil#œj
+^aient$#ɛ
+^eau#o
+^e(?=['a','â','â','o','ô','ô'])# 
+^et$#ɛ
+^oin#wϵ
+^eu(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#œ
+^eu$#ø
+^eue$#ø
+^ati#asj
+^emment$#amα
+^ente$#αt
+^t(?=['in'])#t
+^ce$#sœ
+^t'#t
+^je$#ʒœ
+^ien#jϵ
+^(eur|oeur)$#œr
+^les$#le
+^(ou|ou$)#u
+^à$#a
+^aise$#ɛz
+^en$#α
+^gu#g
+^g(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#g
+^g(?=['o','ô','a','â','u','û'])#g
+^g(?=['e','ë','ê','i','y'])#ʒ
+^ille#ij
+^oi#wa
+^t(?=['i'])#s
+^aine$#ɛn
+^ys#iz
+^an(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#an
+^onne$#ɔn
+^on$#σ
+^oin$#wϵ
+^t$# 
+^p$# 
+^(l|ll)#l
+^j'#ʒ
+^h(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])# 
+^l'#l
+^c'#s
+^sym(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#sϵ
+^y(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#i
+^es$#
+^th#t
+^ine$#in
+^ph#f
+^i(?=['a','â','i','o','ô','u','û'])#j
+^and$#α
+^euse$#øz
+^est$#ɛ
+^aille#aj
+^eille#ɛj
+^s(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#s
+^c(?=['b', 'c', 'd', 'f', 'g', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#k
+^c(?=['a','â','u','û','o','ô'])#k
+^c(?=['e','ë','ê','é','è','y','i','î'])#s
+^ç(?=['a','â','u','û','o','ô'])#s
+^s(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#s
+^s(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#z
+^y(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#j
+^ment$#mα
+^e(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z']{2})#ɛ
+^e(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z']{1})#œ
+^an(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#an
+^en(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#en
+^en(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#α
+^qu#k
+^eu$#œ
+^ent$#
+^ouai#wɛ
+^ouet#wɛ
+^ch(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#k
+^in#ϵ
+^ou(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#w
+^ain#ϵ
+^ein#ϵ
+^er$#e
+^eu$#ø
+^gno#gno
+^hou#u
+^in$#ϵ
+^un$#π
+^ai#ɛ
+^an#α
+^au#o
+^ce$#sə
+^ce#sɛ
+^ch(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#ʃ
+^eu#œ
+^ez#e
+^ff#f
+^gn#ɲ
+^im#ϵ
+^mm#m
+^nn#n
+^on(?=["a","â","e","ë","ê","i","o","ô","u","û","û","y","é","è","â","à","ï","î","ë","û","ù","ê"])#on
+^on(?=['b', 'c', 'd', 'f', 'g', 'h', 'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'v', 'w', 'x', 'z'])#σ
+^om(?=['b'])#σ
+^oû#u
+^pp#p
+^rr#r
+^s$#
+^sh#ʃ
+^ss#s
+^t$#
+^e$#
+^tt#t
+^ui#µi
+^x$#
+^x#ks
+^ça#sa
+^a#a
+^b#b
+^d#d
+^f#f
+^g#ʒ
+^i#i
+^j#ʒ
+^k#k
+^m#m
+^n#n
+^o#ɔ
+^p#p
+^r#r
+^t#t
+^u#y
+^v#v
+^w#w
+^z#z
+^â#a
+^è#ɛ
+^é#e
+^î#i
+^ï#i
+^ô#o
+^ö#o
+^c$#k
+^é#e
+^è#ɛ
+^ê#ɛ
+^ë#œ
+^û#y
+^à#a
+^h#
+^y#i

data/data/exceptions.csv

@@ -0,0 +1,50 @@
+ses#se
+ces#se
+leur#lœr
+n'ai#nɛ
+n'est#nɛ
+n'es#nɛ
+ne#nœ
+es#ɛ
+des#de
+eu#y
+l'#l
+de#dœ
+mon#mσ
+son#sσ
+sur#syr
+sûr#syr
+je#ʒœ
+tu#ty
+il#il
+nous#nu
+vous#vu
+ils#il
+êtes#ɛt
+sont#sσ
+sommes#sɔm
+avez#ave
+du#dy
+le#lœ
+la#la
+les#le
+sans#sα
+moins#mwϵ
+plus#plys
+moi#mwa
+toi#twa
+soi#swa
+eux#ø
+nos#no
+vos#vo
+leurs#lœr
+lui#lµi
+qui#ki
+que#kœ
+quoi#kwa
+donc#dσk
+dont#dσ
+comment#kɔmα
+se#sœ
+me#mœ
+te#tœ