english

dicts/dicts.go

@@ -115,7 +115,7 @@ func GetDict(lang, filename string) ([]byte, error) {
 	case "Farsi":
 		return farsi.Language.ReadFile(filename)
 	case "English":
-		return english.Language.ReadFile(filename)
+		return english.Language.ReadFile(lzw(filename))
 	case "German":
 		return german.Language.ReadFile(filename)
 	case "French":

dicts/english/language.go

@@ -2,5 +2,5 @@ package english
 
 import "embed"
 
-//go:embed *.tsv language.json weights0.json.gz
+//go:embed *.tsv language.json weights1.json.lzw
 var Language embed.FS

dicts/english/language.json

@@ -1,135 +1,80 @@
-{
-	"Map": {"'s":["z","s","ɪz"],
-"a":["æ","ə","eɪ","ɑ","ɛ","ɔ","ʌ","ɚ"],
-"ae":["eɪ"],
-"ah":["ə"],
-"al":["əl","ɔ","ɑ","æ"],
-"all":["əl"],
-"an":["ən","æ","ɑ"],
-"ane":["ən"],
-"ann":["ən","æ"],
-"ao":["aʊ"],
-"ar":["ɚ","ɛ","ɑ"],
-"au":["ɔ","aʊ"],
-"aw":["ɔ"],
-"ay":["eɪ"],
-"aye":["eɪ"],
-"b":["b"],
-"bb":["b"],
-"bbe":["b"],
-"be":["b"],
-"c":["k","s","tʃ"],
-"ce":["s"],
-"ch":["tʃ","k"],
-"che":["tʃ"],
-"ck":["k"],
-"cke":["k"],
-"cs":["ks"],
-"d":["d","dʒ"],
-"dd":["d"],
-"de":["d","n"],
-"dge":["dʒ"],
-"e":["ɛ","i","ɹ","ɪ","ə","aɪ"],
-"ee":["i"],
-"er":["ɚ"],
-"ey":["i"],
-"f":["f"],
-"fe":["f"],
-"ff":["f"],
-"ffe":["f"],
-"g":["g","ŋ","dʒ"],
-"ge":["dʒ"],
-"gg":["g"],
-"gge":["g"],
-"gh":["g"],
-"gs":["gz"],
-"gue":["g"],
-"h":["h"],
-"i":["ɪ","aɪ","i","ɹ"],
-"ia":["iə","i"],
-"ie":["i"],
-"in":["ɪ","ɪn","aɪ","i"],
-"ine":["ɪn"],
-"inn":["ɪn"],
-"ir":["ɹ","ɪ"],
-"ire":["ɹ"],
-"j":["dʒ"],
-"k":["k","ŋ"],
-"ke":["k"],
-"l":["l","ɛ"],
-"le":["əl","l","li"],
-"lee":["li"],
-"ley":["li"],
-"li":["lɪ","li","l"],
-"lie":["li"],
-"ll":["l","ɛ"],
-"lle":["l"],
-"ly":["li"],
-"m":["m"],
-"mb":["m"],
-"me":["m"],
-"mm":["m"],
-"mme":["m"],
-"n":["n","ən","ɛn","ŋ"],
-"ne":["n"],
-"ng":["ŋ","ɪ"],
-"nn":["n"],
-"o":["oʊ","ɑ","ɔ","ə","ʌ","u","ʊ","aʊ"],
-"oe":["oʊ"],
-"oh":["oʊ"],
-"om":["əm","ɑ","oʊ"],
-"omb":["əm"],
-"on":["ən","ɑ","oʊ"],
-"oo":["u"],
-"or":["ɚ","ɔ"],
-"ou":["aʊ","u"],
-"ow":["oʊ"],
-"owe":["oʊ"],
-"p":["p"],
-"pe":["p"],
-"pp":["p"],
-"ppe":["p"],
-"q":["k"],
+{"Map":{"":[],
+"'":["ʊ"],
+"'s":["z","s","ʊ","ɪ"],
+"a":["ə","ˈæ","ˈɑ","æ","ɑ","ˌæ","ˈe","ɪ","ˈɛ","e","ˌɑ","ɔ","ˌe","ˈɔ","ɑː","t","ˌɛ","ʊ","ɛ","ˈi","ˈa","a","ɒ","ˌa","d","ˌɔ","l","ɹ"],
+"ar":["ɑː","ɚˈ","ˈɑɹ","ˈɛ","ɚ","ɑɹ","ɑɹˈ","ˈɛɹ","ˈɑ","ˌɑɹ","ɑ","æɹ","e","ə"],
+"are":["ˈɛɹ"],
+"au":["ˈɔ","ɔ","ɔː","ˈaʊ"],
+"aw":["ɔː"],
+"b":["b","ˈb","ə","bˌ"],
+"be":["ɪb"],
+"c":["k","s","ə","ˈk","ʃ","ˌk","ˈs","kˈ","j","jˈ","kˌ","t"],
+"cc":["tʃ","k"],
+"ce":["ɪs","s","sˈɛ","n"],
+"ch":["tʃ","k","ʃ","ˈk"],
+"cz":["tʃ"],
+"d":["d","ˈd","ɪ","ə","dˌ","dˈ"],
+"e":["ˈɛ","ɪ","ə","i","ɛ","ˌɛ","e","ˈi","ˈa","ˈe","a","ˌi","iː","d","ˈɪ","ˌe","ʊ","ˌa","ɚ","s","jˈ","z"],
+"ea":["ˈɪ","ˈi","iː","ˈɛ","ɛ","i","ˌi","e"],
+"ed":["t","ɪ","d","ˈɛ","e"],
+"er":["ɚ","ˈɚ","ə","ɹ","ˈɛ","ɜː","g","ɪ","d"],
+"ew":["ˈu"],
+"f":["f","ˈf","jˈ","fˌ"],
+"g":["g","dʒ","ˈg","ʒ","ɪ","ə"],
+"ge":["ʒ","g","d"],
+"gh":["ɪ","g"],
+"h":["hˈ","h","ˈh","a","hˌ","ˌo","ɛ"],
+"i":["ɪ","i","ˈɪ","ə","ˌɪ","ˈi","ˈa","a","j","ˌa","ˌi","iː","aɪˈ","iˈ","ɪˈ","s","d","t","ŋ"],
+"ing":["tɪ"],
+"ir":["ˈɚ","ɹ","a","ˈa","ɪ","ɪɹˈ"],
+"j":["dʒˈ","dʒ","j","ˈdʒ"],
+"k":["k","ˈk","ə"],
+"ke":["ɪk","k","ə","kˈɛ","kˈ"],
+"ked":["t"],
+"l":["l","ˈl","ˌa","lˈ","j","a","o","ˈɛ","ɪ","ə","ʊlˈ","lˌ","ɹ"],
+"le":["əl","l","ˈɛ","ɪ","ə","g"],
+"m":["m","ˈm","mˌ","ə","ˈɛ","ˌm","mˈ","ʊmˈ","jˈ","j"],
+"me":["ʊm","ɪm","m","mˈɛ","ə","mˈ","ˈme"],
+"n":["n","ŋ","ˈn","ə","nˈ","ɪ"],
+"ne":["ɪn","ʊn","n","nˈi","nˈɛ","nˈ"],
+"nel":["nˈɛ"],
+"ness":["nə"],
+"net":["nˈɛ"],
+"new":["nˈu"],
+"ng":["ŋ","g","n","t"],
+"o":["ə","ˈɑ","ˈɔ","o","ˈoʊ","ɔ","ʊ","ɑ","ɒ","ˌɑ","ˈa","ˌɔ","ˈʌ","a","ʌ","ʊˈ","ˈɒ","ˌa","ˌo","d"],
+"oa":["ˌoʊ","ˈoʊ"],
+"on":["n","ə","ənˈ","ˈɑn","ən","ˈo","ɒn","o","ɑ","ˌɑn"],
+"on's":["ən"],
+"ons":["ən"],
+"oo":["ˈu","ˈʊ","uː","ˌʊ","ʊ","u","ˌu"],
+"or":["ɚ","ˈɔɹ","ɔː","ˌɔɹ","ɔɹ","ɔɹˈ","ə","ˌɚ","t"],
+"ore":["ɔɹ"],
+"ors":["tɚ"],
+"ou":["ˈaʊ","ə","u","ˈu","aʊ","ˈɔ","ʊ","uː","ˈʌ","ʌ"],
+"p":["p","ˈp","ˌp","ə"],
+"ph":["f"],
+"qu":["kwˈ","kw"],
 "que":["k"],
-"r":["ɹ","ɚ","ɛ"],
-"re":["ɹ","ɚ"],
-"s":["s","z","ʒ","ʃ"],
-"sch":["ʃ"],
-"se":["s","z"],
-"sh":["ʃ"],
-"ss":["s"],
-"sse":["s"],
-"t":["t"],
-"tch":["tʃ"],
-"te":["t"],
-"th":["θ"],
-"tt":["t"],
-"tte":["t"],
-"ty":["ti"],
-"u":["ʌ","u","ju","w","ʊ"],
-"ue":["u"],
-"uer":["ɚ"],
-"ur":["ɚ"],
-"ure":["ɚ"],
-"urr":["ɚ"],
-"v":["v"],
-"ve":["v"],
-"w":["w"],
-"x":["ks","gz"],
-"y":["i","ɪ","aɪ","j"],
-"ye":["aɪ"],
-"z":["z","s"],
-"ze":["z"],
-"é":["eɪ"],
-"ó":["ɔ"]}
-,
-	"SrcMulti": [],
-	"DstMulti": [],
-	"SrcMultiSuffix": [],
-	"DstMultiSuffix": ["ː"],
-	"PrePhonWordSteps": [
-		{"Trim": ".,"},
-		{"ToLower": true}
-	]
-}
-
+"r":["ɹ","ɚ","ˈa","ˈɹ","ˈɚ","ɹˌ","ə","ɹˈ","ˈɛ","a","l"],
+"s":["s","z","ˈs","ə","ʒ","ʊ","ɪ","ʃ","n","t","ˈz","ˌs"],
+"sch":["ʃˈ","ʃ"],
+"sh":["ʃ","ˈʃ"],
+"ss":["ʃ","s","sˈ","ə"],
+"t":["t","ʃ","ˈt","ə","tˈ","ɪ","a","ˌo","ˌa","tˌ","n","ˈɛ","s","ʊ","ˌt"],
+"te":["ɪt","t","ˈɛ","ə","ɪ","ˈte","ˌɛ"],
+"th":["θ","ð","ˈθ"],
+"ts":["nt"],
+"u":["ˈʌ","ə","ʌ","ˈu","u","ˌʌ","uː","juː","ˌu","ˈʊ","wˈ","ʊ","jʊ","j"],
+"ur":["ˈɚ","ɚ","ʃ"],
+"v":["v","ˈv"],
+"ve":["ɪv","v","vˈɛ","ˌov","vˈ"],
+"w":["wˈ","w","ʊ","f","ˈw","a","wˌ"],
+"wr":["ɹˈ"],
+"x":["ks","gzˈ"],
+"y":["i","ɪ","ˈɪ","jˈ","a","ˈa","j","ɹ","ˌa","ˌɪ"],
+"z":["z","s","t","ə"],
+"ze":["ɪz"],
+"zed":["ɪz"]},"SrcMulti":null,"DstMulti":null,"SrcMultiSuffix":null,"DstMultiSuffix":["ː"],
+"DstMultiPrefix":["ˈ","'","ˌ"],
+"DropLast":null,"SplitBefore":null,"SplitAfter":null,"PrePhonWordSteps":[{"Trim":".,"},{"ToLower":true}]}