diff --git a/commands/natural_language.py b/commands/natural_language.py
index cb4a4800..24ff7cfc 100644
--- a/commands/natural_language.py
+++ b/commands/natural_language.py
@@ -17,8 +17,8 @@ __registry__ = cr = CommandRegistry(init_func=_init)
 
 @cr.register('process')
 @cr.restrict(full_command_only=True)
-def process(sentence, ctx_msg, internal=False):
-    sentence = sentence.strip()
+def process(args_text, ctx_msg, internal=False):
+    sentence = args_text.strip()
     potential_commands = parse_potential_commands(sentence)
     potential_commands = sorted(filter(lambda x: x[0] > 60, potential_commands), key=lambda x: x[0], reverse=True)
     if len(potential_commands) > 0:
@@ -26,7 +26,13 @@ def process(sentence, ctx_msg, internal=False):
         ctx_msg['parsed_data'] = most_possible_cmd[3]
         cmdhub.call(most_possible_cmd[1], most_possible_cmd[2], ctx_msg)
     else:
-        core.echo('我暂时不理解你在说什么哦～', ctx_msg, internal)
+        if ctx_msg.get('from_voice'):
+            core.echo('暂时无法理解你的意思，下面将发送图灵机器人的回复……', ctx_msg, internal)
+            core.tuling123(sentence, ctx_msg, internal)
+        else:
+            core.echo('暂时无法理解你的意思。\n'
+                      '由于自然语言识别还非常不完善，建议使用命令来精确控制我。\n'
+                      '如需帮助请发送「使用帮助」。', ctx_msg, internal)
 
 
 def _load_processors():
diff --git a/nl_processor.py b/nl_processor.py
index d20073bf..72a26daa 100644
--- a/nl_processor.py
+++ b/nl_processor.py
@@ -26,7 +26,9 @@ def parse_potential_commands(sentence):
         for regex in processor[0]:
             for word, flag in segmentation:
                 if re.match(regex, word):
-                    potential_commands.append(processor[1](sentence, segmentation))
+                    result = processor[1](sentence, segmentation)
+                    if result:
+                        potential_commands.append(result)
                     processed = True
                     # A word matched, skip the rest of words
                     break
@@ -34,5 +36,7 @@ def parse_potential_commands(sentence):
                 # Current processor has processed, skip the rest of keywords
                 break
     for func in _processors_without_keyword:
-        potential_commands.append(func(sentence, segmentation))
+        result = func(sentence, segmentation)
+        if result:
+            potential_commands.append(result)
     return potential_commands
diff --git a/nl_processors/translate.py b/nl_processors/translate.py
index b3888ded..d6203729 100644
--- a/nl_processors/translate.py
+++ b/nl_processors/translate.py
@@ -3,14 +3,14 @@ import re
 from nl_processor import as_processor
 
 _query_lang_matcher = [
-    re.compile('[把将]?[ ,.，。]?(.*?)[ ,.，。]?(?:这[个]?(?:词[组]?|句(?:子|话)?|短语))翻译[成为到](\w+?[文语])(?![ :：,，.。])'),
-    re.compile('(\w+?)[ ,.，。]?(?:这[个]?(?:词[组]?|句(?:子|话)?|短语))?[的用](\w+?[文语])')
+    re.compile('[把将]?[\s,.，。]?(.*?)[\s,.，。]?(?:这[个]?(?:词[组]?|句(?:子|话)?|短语))?翻译[成为到](\w+?[文语])(?![\s:：,，.。])'),
+    re.compile('(\w+?)[\s,.，。]?(?:这[个]?(?:词[组]?|句(?:子|话)?|短语))?[的用](\w+?[文语])')
 ]
 
 _lang_query_matcher = [
-    re.compile('[把将]?(?:(?:这[个]?|[下后][面]?)(?:词[组]?|句(?:子|话)?|短语))翻译[成为到](\w+?[文语])[ :：,，.。](.*)'),
+    re.compile('.*[把将]?(?:(?:这[个]?|[下后][面]?)(?:词[组]?|句(?:子|话)?|短语))?翻译[成为到]\s*(\w+?[文语])[\s:：,，](.*)'),
     re.compile('[用]?(\w+[文语])\w+?(?:说|讲|表达|表示)(.*)(?:这[个]?(?:词[组]?|句(?:子|话)?|短语))'),
-    re.compile('[用]?(\w+[文语])\w+?(?:说|讲|表达|表示)(.*)')
+    re.compile('[用]?(\w+[文语])\w+?(?:说|讲|表达|表示)(.*)'),
 ]
 
 
@@ -27,5 +27,6 @@ def _processor(sentence, segmentation):
                 lang, query = m.group(2), m.group(1)
             break
     if lang and query:
+        print('翻译: 目标语言:', lang, ', 待翻译文本:', query)
         return 90, 'translate.translate_to', ' '.join((lang.strip(), query.strip(' ,，'))), None
     return None