Replaced Encodec with Vocos

app.py

@@ -36,6 +36,8 @@ import gradio as gr
 from vocos import Vocos
 from transformers import WhisperProcessor, WhisperForConditionalGeneration
 
+from utils.sentence_cutter import split_text_into_sentences
+
 torch._C._jit_set_profiling_executor(False)
 torch._C._jit_set_profiling_mode(False)
 torch._C._set_graph_executor_optimize(False)
@@ -321,7 +323,7 @@ def infer_from_prompt(text, language, accent, preset_prompt, prompt_file):
     return message, (24000, samples.squeeze(0).cpu().numpy())
 
 
-from utils.sentence_cutter import split_text_into_sentences
+
 @torch.no_grad()
 def infer_long_text(text, preset_prompt, prompt=None, language='auto', accent='no-accent'):
     """

utils/g2p/english.py

@@ -19,6 +19,7 @@ hyperparameter. Some cleaners are English-specific. You'll typically want to use
 import re
 from unidecode import unidecode
 import inflect
+import eng_to_ipa as ipa
 _inflect = inflect.engine()
 _comma_number_re = re.compile(r'([0-9][0-9\,]+[0-9])')
 _decimal_number_re = re.compile(r'([0-9]+\.[0-9]+)')
@@ -157,7 +158,6 @@ def mark_dark_l(text):
 
 
 def english_to_ipa(text):
-    import eng_to_ipa as ipa
     text = unidecode(text).lower()
     text = expand_abbreviations(text)
     text = normalize_numbers(text)

utils/g2p/japanese.py

@@ -1,5 +1,6 @@
 import re
 from unidecode import unidecode
+import pyopenjtalk
 
 
 
@@ -73,7 +74,7 @@ def symbols_to_japanese(text):
 
 def japanese_to_romaji_with_accent(text):
     '''Reference https://r9y9.github.io/ttslearn/latest/notebooks/ch10_Recipe-Tacotron.html'''
-    import pyopenjtalk
+
     text = symbols_to_japanese(text)
     sentences = re.split(_japanese_marks, text)
     marks = re.findall(_japanese_marks, text)

utils/g2p/mandarin.py

@@ -4,6 +4,7 @@ import re
 import jieba
 import cn2an
 import logging
+from pypinyin import lazy_pinyin, BOPOMOFO
 
 
 # List of (Latin alphabet, bopomofo) pairs:
@@ -240,7 +241,7 @@ def number_to_chinese(text):
 
 
 def chinese_to_bopomofo(text):
-    from pypinyin import lazy_pinyin, BOPOMOFO
+
     text = text.replace('、', '，').replace('；', '，').replace('：', '，')
     words = jieba.lcut(text, cut_all=False)
     text = ''