2nd version of the tokenizer fix

2023-10-06 17:39:04 +02:00 · 2023-10-06 17:39:04 +02:00 · 2fdf51ebd2
parent 1ec341857e
commit 2fdf51ebd2
1 changed files with 5 additions and 3 deletions
--- a/TTS/tts/layers/xtts/tokenizer.py
+++ b/TTS/tts/layers/xtts/tokenizer.py
@ -223,9 +223,11 @@ class VoiceBpeTokenizer:
            results = kks.convert(txt)
            txt = " ".join([result["kana"] for result in results])
            txt = basic_cleaners(txt)
-        # elif lang == "en":
-        #     txt = english_cleaners(txt)
-        # English cleaner remove the language tag [en]
+        elif lang == "en":
+            if txt[:4] == "[en]":
+                txt = txt[4:]
+            txt = english_cleaners(txt)
+            txt = "[en]" + txt
        elif lang == "ar":
            txt = arabic_cleaners(txt)
        elif lang == "zh-cn":