Adding Catalan (4 accents)

data/ca-ba/language.yml

@@ -0,0 +1,50 @@
+---
+
+language:
+  name: "Balear Catalan"
+  code: "ca-ba"
+  phonemes: !env "${config_dir}/phonemes.txt"
+  keep_stress: true
+
+lexicon: !env "${config_dir}/lexicon.db"
+
+g2p:
+  model: !env "${config_dir}/g2p.fst"
+
+symbols:
+  casing: "lower"
+  number_regex: "^-?\\d+([,.]\\d+)*$"
+  token_split: "\\s+"
+  token_join: " "
+  minor_breaks:
+    - ","
+    - ":"
+    - ";"
+    - "..."
+  major_breaks:
+    - "."
+    - "?"
+    - "!"
+  replace:
+    "[\\<\\>\\(\\)\\[\\]\"]+": ""
+    "\\B'": "\""
+    "'\\B": "\""
+    "’": "'"
+    "'": ""
+    "-": ""
+    "l·l": "l"
+  punctuations:
+    - "\""
+    - "„"
+    - "“"
+    - "”"
+    - "«"
+    - "»"
+    - ","
+    - ":"
+    - ";"
+    - "."
+    - "?"
+    - "¿" 
+    - "!"
+    - "¡"

data/ca-ba/phonemes.txt

@@ -0,0 +1,44 @@
+# https://en.wikipedia.org/wiki/Catalan_phonology
+# Catalan phonemes
+
+p [p]ala
+b [b]ala
+t [t]ela
+d [d]onar
+k [k]ala
+ɡ [g]ala
+m [m]ala
+ɲ fa[ng]
+β aca[b]a
+ð ca[d]a
+ɣ ama[g]ar
+f [f]als
+v a[f]ganès
+s [s]ala
+z ca[s]a
+ʃ [x]oc
+ʒ mà[g]ic
+tʃ co[tx]e
+dʒ me[tg]e
+l [l]íquid
+ʎ [ll]amp
+r ca[rr]o
+ɾ ca[r]a
+w ve[u]en
+uw ca[u]re
+j ca[i]re
+y [i]a[i]a
+n [n]ena
+ŋ pi[n]güí
+ts po[ts]er
+dz do[tz]e
+
+# Vowels
+i r[i]c
+e c[e]c
+ɛ s[e]c
+a s[a]c
+ɔ f[o]c
+o s[ó]c
+u s[u]c
+ə [a]mor

data/ca-ce/language.yml

@@ -0,0 +1,50 @@
+---
+
+language:
+  name: "Central Catalan"
+  code: "ca-ce"
+  phonemes: !env "${config_dir}/phonemes.txt"
+  keep_stress: true
+
+lexicon: !env "${config_dir}/lexicon.db"
+
+g2p:
+  model: !env "${config_dir}/g2p.fst"
+
+symbols:
+  casing: "lower"
+  number_regex: "^-?\\d+([,.]\\d+)*$"
+  token_split: "\\s+"
+  token_join: " "
+  minor_breaks:
+    - ","
+    - ":"
+    - ";"
+    - "..."
+  major_breaks:
+    - "."
+    - "?"
+    - "!"
+  replace:
+    "[\\<\\>\\(\\)\\[\\]\"]+": ""
+    "\\B'": "\""
+    "'\\B": "\""
+    "’": "'"
+    "'": ""
+    "-": ""
+    "l·l": "l"
+  punctuations:
+    - "\""
+    - "„"
+    - "“"
+    - "”"
+    - "«"
+    - "»"
+    - ","
+    - ":"
+    - ";"
+    - "."
+    - "?"
+    - "¿" 
+    - "!"
+    - "¡"

data/ca-ce/phonemes.txt

@@ -0,0 +1,44 @@
+# https://en.wikipedia.org/wiki/Catalan_phonology
+# Catalan phonemes
+
+p [p]ala
+b [b]ala
+t [t]ela
+d [d]onar
+k [k]ala
+ɡ [g]ala
+m [m]ala
+ɲ fa[ng]
+β aca[b]a
+ð ca[d]a
+ɣ ama[g]ar
+f [f]als
+v a[f]ganès
+s [s]ala
+z ca[s]a
+ʃ [x]oc
+ʒ mà[g]ic
+tʃ co[tx]e
+dʒ me[tg]e
+l [l]íquid
+ʎ [ll]amp
+r ca[rr]o
+ɾ ca[r]a
+w ve[u]en
+uw ca[u]re
+j ca[i]re
+y [i]a[i]a
+n [n]ena
+ŋ pi[n]güí
+ts po[ts]er
+dz do[tz]e
+
+# Vowels
+i r[i]c
+e c[e]c
+ɛ s[e]c
+a s[a]c
+ɔ f[o]c
+o s[ó]c
+u s[u]c
+ə [a]mor

data/ca-no/language.yml

@@ -0,0 +1,50 @@
+---
+
+language:
+  name: "Nord-Occidental Catalan"
+  code: "ca-no"
+  phonemes: !env "${config_dir}/phonemes.txt"
+  keep_stress: true
+
+lexicon: !env "${config_dir}/lexicon.db"
+
+g2p:
+  model: !env "${config_dir}/g2p.fst"
+
+symbols:
+  casing: "lower"
+  number_regex: "^-?\\d+([,.]\\d+)*$"
+  token_split: "\\s+"
+  token_join: " "
+  minor_breaks:
+    - ","
+    - ":"
+    - ";"
+    - "..."
+  major_breaks:
+    - "."
+    - "?"
+    - "!"
+  replace:
+    "[\\<\\>\\(\\)\\[\\]\"]+": ""
+    "\\B'": "\""
+    "'\\B": "\""
+    "’": "'"
+    "'": ""
+    "-": ""
+    "l·l": "l"
+  punctuations:
+    - "\""
+    - "„"
+    - "“"
+    - "”"
+    - "«"
+    - "»"
+    - ","
+    - ":"
+    - ";"
+    - "."
+    - "?"
+    - "¿" 
+    - "!"
+    - "¡"

data/ca-no/phonemes.txt

@@ -0,0 +1,44 @@
+# https://en.wikipedia.org/wiki/Catalan_phonology
+# Catalan phonemes
+
+p [p]ala
+b [b]ala
+t [t]ela
+d [d]onar
+k [k]ala
+ɡ [g]ala
+m [m]ala
+ɲ fa[ng]
+β aca[b]a
+ð ca[d]a
+ɣ ama[g]ar
+f [f]als
+v a[f]ganès
+s [s]ala
+z ca[s]a
+ʃ [x]oc
+ʒ mà[g]ic
+tʃ co[tx]e
+dʒ me[tg]e
+l [l]íquid
+ʎ [ll]amp
+r ca[rr]o
+ɾ ca[r]a
+w ve[u]en
+uw ca[u]re
+j ca[i]re
+y [i]a[i]a
+n [n]ena
+ŋ pi[n]güí
+ts po[ts]er
+dz do[tz]e
+
+# Vowels
+i r[i]c
+e c[e]c
+ɛ s[e]c
+a s[a]c
+ɔ f[o]c
+o s[ó]c
+u s[u]c
+ə [a]mor

data/ca-va/language.yml

@@ -0,0 +1,50 @@
+---
+
+language:
+  name: "Valencià Catalan"
+  code: "ca-va"
+  phonemes: !env "${config_dir}/phonemes.txt"
+  keep_stress: true
+
+lexicon: !env "${config_dir}/lexicon.db"
+
+g2p:
+  model: !env "${config_dir}/g2p.fst"
+
+symbols:
+  casing: "lower"
+  number_regex: "^-?\\d+([,.]\\d+)*$"
+  token_split: "\\s+"
+  token_join: " "
+  minor_breaks:
+    - ","
+    - ":"
+    - ";"
+    - "..."
+  major_breaks:
+    - "."
+    - "?"
+    - "!"
+  replace:
+    "[\\<\\>\\(\\)\\[\\]\"]+": ""
+    "\\B'": "\""
+    "'\\B": "\""
+    "’": "'"
+    "'": ""
+    "-": ""
+    "l·l": "l"
+  punctuations:
+    - "\""
+    - "„"
+    - "“"
+    - "”"
+    - "«"
+    - "»"
+    - ","
+    - ":"
+    - ";"
+    - "."
+    - "?"
+    - "¿" 
+    - "!"
+    - "¡"

data/ca-va/phonemes.txt

@@ -0,0 +1,44 @@
+# https://en.wikipedia.org/wiki/Catalan_phonology
+# Catalan phonemes
+
+p [p]ala
+b [b]ala
+t [t]ela
+d [d]onar
+k [k]ala
+ɡ [g]ala
+m [m]ala
+ɲ fa[ng]
+β aca[b]a
+ð ca[d]a
+ɣ ama[g]ar
+f [f]als
+v a[f]ganès
+s [s]ala
+z ca[s]a
+ʃ [x]oc
+ʒ mà[g]ic
+tʃ co[tx]e
+dʒ me[tg]e
+l [l]íquid
+ʎ [ll]amp
+r ca[rr]o
+ɾ ca[r]a
+w ve[u]en
+uw ca[u]re
+j ca[i]re
+y [i]a[i]a
+n [n]ena
+ŋ pi[n]güí
+ts po[ts]er
+dz do[tz]e
+
+# Vowels
+i r[i]c
+e c[e]c
+ɛ s[e]c
+a s[a]c
+ɔ f[o]c
+o s[ó]c
+u s[u]c
+ə [a]mor

gruut-lang-ca/LANGUAGE

@@ -0,0 +1 @@
+ca-ce Catalan

gruut-lang-ca/README.md

@@ -0,0 +1,3 @@
+# gruut Catalan
+
+Language-specific files for Catalan (ca) in [gruut](https://github.com/rhasspy/gruut)

gruut-lang-ca/gruut_lang_ca/VERSION

@@ -0,0 +1 @@
+0.0.0