fix: recover empty phoneme_id_map for ubl, tzo-chenalhó, ubu from MMS vocab.txt

These 3 dialect voices shipped with empty phoneme_id_map (0 symbols) so every input char was dropped (empty audio). Recovered the exact char->id mapping from Meta MMS per-language vocab.txt (dl.fbaipublicfiles.com/mms/tts/<iso>.tar.gz), verified vocab size == ONNX emb table (ubl 37, chenalhó 36, ubu 28) and 100% token survival on synth.

facebook__mms-tts-tzo-dialect_chenalhó-Tzotzil/config.json

@@ -7,7 +7,7 @@
   "audio": {
     "sample_rate": 16000
   },
-  "num_symbols": 256,
+  "num_symbols": 36,
   "num_speakers": 1,
   "num_langs": 1,
   "speaker_id_map": {},
@@ -19,14 +19,51 @@
     "length_scale": 1.0,
     "noise_w": 0.8
   },
-  "phoneme_id_map": {},
+  "phoneme_id_map": {
+    "|": 0,
+    "a": 1,
+    "c": 2,
+    "i": 3,
+    "l": 4,
+    "t": 5,
+    "'": 6,
+    "u": 7,
+    "o": 8,
+    "e": 9,
+    "n": 10,
+    "s": 11,
+    "j": 12,
+    "h": 13,
+    "b": 14,
+    "y": 15,
+    "x": 16,
+    "m": 17,
+    "v": 18,
+    "q": 19,
+    "p": 20,
+    "r": 21,
+    "d": 22,
+    "—": 23,
+    "g": 24,
+    "z": 25,
+    "-": 26,
+    "í": 27,
+    "f": 28,
+    "ó": 29,
+    "é": 30,
+    "ú": 31,
+    "á": 32,
+    "k": 33,
+    "ñ": 34,
+    " ": 35
+  },
   "pad": null,
-  "blank": null,
+  "blank": "|",
   "bos": null,
   "eos": null,
   "add_blank_char": true,
   "add_blank_word": false,
-  "use_eos_bos": true,
+  "use_eos_bos": false,
   "blank_at_start": true,
   "blank_at_end": true,
   "word_sep_token": " ",

facebook__mms-tts-ubl-Bikol_Buhi’non/config.json

@@ -7,7 +7,7 @@
   "audio": {
     "sample_rate": 16000
   },
-  "num_symbols": 256,
+  "num_symbols": 37,
   "num_speakers": 1,
   "num_langs": 1,
   "speaker_id_map": {},
@@ -19,14 +19,52 @@
     "length_scale": 1.0,
     "noise_w": 0.8
   },
-  "phoneme_id_map": {},
+  "phoneme_id_map": {
+    "j": 0,
+    "a": 1,
+    "_": 2,
+    "'": 3,
+    "k": 4,
+    "e": 5,
+    "w": 6,
+    "i": 7,
+    "f": 8,
+    "u": 9,
+    "o": 10,
+    "d": 11,
+    "n": 12,
+    "l": 13,
+    "5": 14,
+    "h": 15,
+    " ": 16,
+    "-": 17,
+    "v": 18,
+    "4": 19,
+    "6": 20,
+    "p": 21,
+    "t": 22,
+    "3": 23,
+    "g": 24,
+    "c": 25,
+    "0": 26,
+    "b": 27,
+    "m": 28,
+    "2": 29,
+    "s": 30,
+    "z": 31,
+    "x": 32,
+    "q": 33,
+    "r": 34,
+    "1": 35,
+    "y": 36
+  },
   "pad": null,
-  "blank": null,
+  "blank": "j",
   "bos": null,
   "eos": null,
   "add_blank_char": true,
   "add_blank_word": false,
-  "use_eos_bos": true,
+  "use_eos_bos": false,
   "blank_at_start": true,
   "blank_at_end": true,
   "word_sep_token": " ",

facebook__mms-tts-ubu-Umbu-Ungu/config.json

@@ -7,7 +7,7 @@
   "audio": {
     "sample_rate": 16000
   },
-  "num_symbols": 256,
+  "num_symbols": 28,
   "num_speakers": 1,
   "num_langs": 1,
   "speaker_id_map": {},
@@ -19,14 +19,43 @@
     "length_scale": 1.0,
     "noise_w": 0.8
   },
-  "phoneme_id_map": {},
+  "phoneme_id_map": {
+    "r": 0,
+    " ": 1,
+    "i": 2,
+    "'": 3,
+    "o": 4,
+    "ú": 5,
+    "n": 6,
+    "t": 7,
+    "p": 8,
+    "s": 9,
+    "e": 10,
+    "ó": 11,
+    "í": 12,
+    "m": 13,
+    "w": 14,
+    "a": 15,
+    "u": 16,
+    "_": 17,
+    "á": 18,
+    "b": 19,
+    "g": 20,
+    "j": 21,
+    "l": 22,
+    "y": 23,
+    "-": 24,
+    "k": 25,
+    "d": 26,
+    "é": 27
+  },
   "pad": null,
-  "blank": null,
+  "blank": "r",
   "bos": null,
   "eos": null,
   "add_blank_char": true,
   "add_blank_word": false,
-  "use_eos_bos": true,
+  "use_eos_bos": false,
   "blank_at_start": true,
   "blank_at_end": true,
   "word_sep_token": " ",