Merge pull request stakira#1082 from Cadlaxa/cadlaxa-local

Add support for External Ending notes via `-` or `R` + proper test file

Author: stakira

OpenUtau.Plugin.Builtin/ArpasingPlusPhonemizer.cs

@@ -46,8 +46,9 @@ public class ArpasingPlusPhonemizer : SyllableBasedPhonemizer {
         protected override string GetDictionaryName() => "cmudict-0_7b.txt";
         protected override Dictionary<string, string> GetDictionaryPhonemesReplacement() => dictionaryReplacements;
 
+
         // For banks with missing vowels
-        private readonly Dictionary<string, string> missingVphonemes = "ax=ah,aa=ah,ae=ah,iy=ih,uh=uw,ix=ih,ux=uh,oh=ao,eu=uh,oe=ax,uy=uw,yw=uw,yx=iy,wx=uw,ea=eh,ia=iy,oa=ao,ua=uw".Split(',')
+        private readonly Dictionary<string, string> missingVphonemes = "ax=ah,aa=ah,ae=ah,iy=ih,uh=uw,ix=ih,ux=uh,oh=ao,eu=uh,oe=ax,uy=uw,yw=uw,yx=iy,wx=uw,ea=eh,ia=iy,oa=ao,ua=uw,R=-".Split(',')
                 .Select(entry => entry.Split('='))
                 .Where(parts => parts.Length == 2)
                 .Where(parts => parts[0] != parts[1])
@@ -97,9 +98,13 @@ public class ArpasingPlusPhonemizer : SyllableBasedPhonemizer {
         private readonly string[] ccvException = { "ch", "dh", "dx", "fh", "gh", "hh", "jh", "kh", "ph", "ng", "sh", "th", "vh", "wh", "zh" };
         private readonly string[] vc_cAcception = { "r", "l" };
         private readonly string[] RomajiException = { "a", "e", "i", "o", "u" };
+        private string[] tails = "-,R,RB".Split(',');
 
         protected override string[] GetSymbols(Note note) {
             string[] original = base.GetSymbols(note);
+            if (tails.Contains(note.lyric)) {
+                return new string[] { note.lyric };
+            }
             if (original == null) {
                 return null;
             }
@@ -198,9 +203,10 @@ protected override IG2p LoadBaseDictionary() {
             g2ps.Add(new ArpabetPlusG2p());
             return new G2pFallbacks(g2ps.ToArray());
         }
-        
         protected override List<string> ProcessSyllable(Syllable syllable) {
-            string prevV = syllable.prevV;
+            syllable.prevV = tails.Contains(syllable.prevV) ? "" : syllable.prevV;
+            var prevV = syllable.prevV == "" ? "" : $"{syllable.prevV}";
+            //string prevV = syllable.prevV;
             string[] cc = syllable.cc;
             string v = syllable.v;
             string basePhoneme;
@@ -354,7 +360,7 @@ protected override List<string> ProcessSyllable(Syllable syllable) {
                 for (var i = firstC; i < cc.Length - 1; i++) {
                     var ccv = $"{string.Join("", cc)} {v}";
                     var ccv1 = string.Join("", cc.Skip(i)) + " " + v;
-                    if (syllable.CurrentWordCc.Length >= 2 && !ccvException.Contains(cc[0])) {
+                    if (syllable.CurrentWordCc.Length >= 2 && !ccvException.Contains(cc[i] + cc[i + 1])) {
                         if (HasOto(ccv, syllable.vowelTone) || HasOto(ValidateAlias(ccv), syllable.vowelTone)) {
                             basePhoneme = ccv;
                             lastC = i;
@@ -375,7 +381,7 @@ protected override List<string> ProcessSyllable(Syllable syllable) {
                     var vc = $"{prevV} {cc[0]}";
                     // CCV will trigger VCC
                     bool CCV = false;
-                    if (syllable.CurrentWordCc.Length >= 2 && !ccvException.Contains(cc[0])) {
+                    if (syllable.CurrentWordCc.Length >= 2 && !ccvException.Contains(cc[1])) {
                         if (HasOto($"{string.Join("", cc)} {v}", syllable.vowelTone) || HasOto(ValidateAlias($"{string.Join("", cc)} {v}"), syllable.vowelTone)) {
                             CCV = true;
                         }
@@ -516,6 +522,9 @@ protected override List<string> ProcessEnding(Ending ending) {
             var phonemes = new List<string>();
             var lastC = cc.Length - 1;
             var firstC = 0;
+            if (tails.Contains(ending.prevV)) {
+                return new List<string>();
+            }
             if (ending.IsEndingV) {
                 var vR = $"{v} -";
                 var vR1 = $"{v} R";

OpenUtau.Test/Files/en_arpa-plus/A#3/oto.ini

@@ -9,12 +9,12 @@ subbanks:
   - F6-B7
 - color: ''
   prefix: ''
-  suffix: _1
+  suffix: _A#3
   tone_ranges:
   - D#5-E6
 - color: ''
   prefix: ''
-  suffix: _3
+  suffix: _C3
   tone_ranges:
   - B2-C#4
 - color: Whisper

OpenUtau.Test/Files/en_arpa-plus/C3/oto.ini

@@ -13,15 +13,15 @@ protected override Phonemizer CreatePhonemizer() {
 
         [Theory]
         [InlineData("en_arpa-plus",
-            new string[] { "good", "morning" },
-            new string[] { "C4", "C4" },
+            new string[] { "good", "morning", },
+            new string[] { "A#3", "A#3" },
             new string[] { "", "" },
-            new string[] { "- g_3", "g uh_3", "uh d_3", "d m_3", "m ao_3", "ao r_3", "r n_3", "n ih_3", "ih ng_3", "ng -_3" })]
+            new string[] { "- g_C3", "g uh_C3", "uh d_C3", "d m_C3", "m ao_C3", "ao r_C3", "r n_C3", "n ih_C3", "ih ng_C3", "ng -_C3" })]
         [InlineData("en_arpa-plus",
             new string[] { "good", "morning" },
-            new string[] { "A3", "F4" },
+            new string[] { "C3", "C3" },
             new string[] { "", "" },
-            new string[] { "- g_3", "g uh_3", "uh d_3", "d m_3", "m ao", "ao r", "r n", "n ih", "ih ng", "ng -" })]
+            new string[] { "- g_C3", "g uh_C3", "uh d_C3", "d m_C3", "m ao_C3", "ao r_C3", "r n_C3", "n ih_C3", "ih ng_C3", "ng -_C3" })]
         public void PhonemizeTest(string singerName, string[] lyrics, string[] tones, string[] colors, string[] aliases) {
             RunPhonemizeTest(singerName, lyrics, RepeatString(lyrics.Length, ""), tones, colors, aliases);
         }
@@ -32,7 +32,7 @@ public void ColorTest() {
                 new NoteParams {
                     lyric = "hi",
                     hint = "",
-                    tone = "C4",
+                    tone = "A#3",
                     phonemes = new PhonemeParams[] {
                         new PhonemeParams {
                             alt = 0,
@@ -51,17 +51,41 @@ public void ColorTest() {
                         }
                     }
                 }
-            }, new string[] { "- hh_3", "hh ay_W", "ay -_3" });
+            }, new string[] { "- hh_C3", "hh ay_W", "ay -_C3" });
+        }
+        public void SyllableTest(string lyric, string hint, string[] aliases) {
+            RunPhonemizeTest("en_arpa-plus", new NoteParams[] { new NoteParams { lyric = lyric, hint = hint, tone = "C3", phonemes = SamePhonemeParams(4, 0, 0, "") } }, aliases);
         }
+        [Theory]
+        [InlineData("read", "", new string[] { "- r_C3", "r eh_C3", "eh d_C3", "d -_C3" })]
+        [InlineData("read", "r iy d", new string[] { "- r_C3", "r iy_C3", "iy d_C3", "d -_C3" })]
 
+        [InlineData("asdfjkl", "r iy d", new string[] { "- r_C3", "r iy_C3", "iy d_C3", "d -_C3" })]
+        [InlineData("", "r iy d", new string[] { "- r_C3", "r iy_C3", "iy d_C3", "d -_C3" })]
+
+        public void SyllableExternalEndingTest(string lyric, string hint, string[] aliases) {
+            RunPhonemizeTest("en_arpa-plus", new NoteParams[] { new NoteParams { lyric = lyric, hint = hint, tone = "C3", phonemes = SamePhonemeParams(4, 0, 0, "") } }, aliases);
+        }
+        [Theory]
+        [InlineData("more", "m aor", new string[] { "- m_C3", "m ao_C3", "ao r_C3", "r -_C3" })]
+        [InlineData("'a", "q ax hh", new string[] { "- q_C3", "q ax_C3", "ax hh_C3", "hh -_C3" })]
+
+        public void SyllableCCVTest(string lyric, string hint, string[] aliases) {
+            RunPhonemizeTest("en_arpa-plus", new NoteParams[] { new NoteParams { lyric = lyric, hint = hint, tone = "C3", phonemes = SamePhonemeParams(4, 0, 0, "") } }, aliases);
+        }
+        [Theory]
+        [InlineData("trusting", "", new string[] { "- tr_C3", "tr ah_C3", "ah st_C3", "st ih_C3", "ih ng_C3", "ng -_C3" })]
+        [InlineData("drive", "", new string[] { "- dr_C3", "dr ay_C3", "ay v_C3", "v -_C3" })]
+
+        public void SyllableFallbackTest(string lyric, string hint, string[] aliases) {
+            RunPhonemizeTest("en_arpa-plus", new NoteParams[] { new NoteParams { lyric = lyric, hint = hint, tone = "C3", phonemes = SamePhonemeParams(4, 0, 0, "") } }, aliases);
+        }
         [Theory]
-        [InlineData("read", "", new string[] { "- r_3", "r eh_3", "eh d_3", "d -_3" })]
-        [InlineData("read", "r iy d", new string[] { "- r_3", "r iy_3", "iy d_3", "d -_3" })]
+        [InlineData("kroidroi", "", new string[] { "- kr_C3", "kr oy_C3", "iy dr_C3", "dr oy_C3", "oy -_C3" })]
+        [InlineData("whhat", "",  new string[] { "- hh_C3", "hh uw_C3", "w ah_C3", "ah t_C3", "t -_C3" })]
 
-        [InlineData("asdfjkl", "r iy d", new string[] { "- r_3", "r iy_3", "iy d_3", "d -_3" })]
-        [InlineData("", "r iy d", new string[] { "- r_3", "r iy_3", "iy d_3", "d -_3" })]
         public void HintTest(string lyric, string hint, string[] aliases) {
-            RunPhonemizeTest("en_arpa-plus", new NoteParams[] { new NoteParams { lyric = lyric, hint = hint, tone = "C4", phonemes = SamePhonemeParams(4, 0, 0, "")} }, aliases);
+            RunPhonemizeTest("en_arpa-plus", new NoteParams[] { new NoteParams { lyric = lyric, hint = hint, tone = "C3", phonemes = SamePhonemeParams(4, 0, 0, "")} }, aliases);
         }
     }
 }