update g2pW dict

paddlespeech/resource/pretrained_models.py

@@ -1345,7 +1345,7 @@ g2pw_onnx_models = {
             'url':
             'https://paddlespeech.bj.bcebos.com/Parakeet/released_models/g2p/G2PWModel.tar',
             'md5':
-            '86a3dd8db0291c575c46e134111dce23',
+            '63bc0894af15a5a591e58b2130a2bcac',
         },
     },
 }

paddlespeech/t2s/frontend/g2pw/onnx_api.py

@@ -8,7 +8,7 @@ import onnxruntime
 import numpy as np
 
 from opencc import OpenCC
-
+from pypinyin import pinyin, lazy_pinyin, Style
 from paddlenlp.transformers import BertTokenizer
 from paddlespeech.utils.env import MODEL_HOME
 from paddlespeech.t2s.frontend.g2pw.dataset import prepare_data,\
@@ -127,6 +127,7 @@ class G2PWOnnxConverter:
         }
         texts, query_ids, sent_ids, partial_results = [], [], [], []
         for sent_id, sent in enumerate(sentences):
+            pypinyin_result = pinyin(sent,style=Style.TONE3)
             partial_result = [None] * len(sent)
             for i, char in enumerate(sent):
                 if char in polyphonic_chars:
@@ -136,6 +137,7 @@ class G2PWOnnxConverter:
                 elif char in monophonic_chars_dict:
                     partial_result[i] =  self.style_convert_func(monophonic_chars_dict[char])
                 elif char in self.char_bopomofo_dict:
-                    partial_result[i] =  self.style_convert_func(self.char_bopomofo_dict[char][0])
+                    partial_result[i] =  pypinyin_result[i][0]
+                    # partial_result[i] =  self.style_convert_func(self.char_bopomofo_dict[char][0])
             partial_results.append(partial_result)
         return texts, query_ids, sent_ids, partial_results

paddlespeech/t2s/frontend/polyphonic.yaml

@@ -1,5 +1,26 @@
 polyphonic:
     湖泊: ['hu2','po1']
+    地壳: ['di4','qiao4']
+    柏树: ['bai3','shu4']
+    曝光: ['bao4','guang1']
     弹力: ['tan2','li4']
+    字帖: ['zi4','tie4']
+    口吃: ['kou3','chi1']
+    包扎: ['bao1','za1']
+    哪吒: ['ne2','zha1']
+    说服: ['shuo1','fu2']
+    识字: ['shi2','zi4']
+    骨头: ['gu3','tou5']
+    对称: ['dui4','chen4']
+    口供: ['kou3','gong4']
+    抹布: ['ma1','bu4']
+    露背: ['lu4','bei4']
+    圈养: ['juan4', 'yang3']
+    眼眶: ['yan3', 'kuang4']
+    品行: ['pin3','xing2']
     颤抖: ['chan4','dou3']
+    差不多: ['cha4','bu5','duo1']
     鸭绿江: ['ya1','lu4','jiang1']
+    撒切尔: ['sa4','qie4','er3']
+    比比皆是: ['bi3','bi3','jie1','shi4']
+    身无长物: ['shen1','wu2','chang2','wu4']