diff --git a/metagpt/actions/azure_tts.py b/metagpt/actions/azure_tts.py
index f528ba001..3520de8b4 100644
--- a/metagpt/actions/azure_tts.py
+++ b/metagpt/actions/azure_tts.py
@@ -4,11 +4,13 @@
 @Time    : 2023/6/9 22:22
 @Author  : Leo Xiao
 @File    : azure_tts.py
+@Modified By: mashenquan, 2023-8-9, add more text formatting options
 """
 from azure.cognitiveservices.speech import AudioConfig, SpeechConfig, SpeechSynthesizer
 
 from metagpt.actions.action import Action
 from metagpt.config import Config
+from metagpt.const import WORKSPACE_ROOT
 
 
 class AzureTTS(Action):
@@ -17,7 +19,7 @@ class AzureTTS(Action):
         self.config = Config()
 
     # 参数参考：https://learn.microsoft.com/zh-cn/azure/cognitive-services/speech-service/language-support?tabs=tts#voice-styles-and-roles
-    def synthesize_speech(self, lang, voice, role, text, output_file):
+    def synthesize_speech(self, lang, voice, text, output_file):
         subscription_key = self.config.get('AZURE_TTS_SUBSCRIPTION_KEY')
         region = self.config.get('AZURE_TTS_REGION')
         speech_config = SpeechConfig(
@@ -29,25 +31,47 @@ class AzureTTS(Action):
             speech_config=speech_config,
             audio_config=audio_config)
 
-        # if voice=="zh-CN-YunxiNeural":
+        # More detail: https://learn.microsoft.com/en-us/azure/ai-services/speech-service/speech-synthesis-markup-voice
         ssml_string = f"""
             <speak version='1.0' xmlns='http://www.w3.org/2001/10/synthesis' xml:lang='{lang}' xmlns:mstts='http://www.w3.org/2001/mstts'>
                 <voice name='{voice}'>
-                    <mstts:express-as style='affectionate' role='{role}'>
-                        {text}
-                    </mstts:express-as>
+                    {text}
                 </voice>
             </speak>
             """
 
-        synthesizer.speak_ssml_async(ssml_string).get()
+        return synthesizer.speak_ssml_async(ssml_string).get()
 
+    @staticmethod
+    def role_style_text(role, style, text):
+        return f'<mstts:express-as role="{role}" style="{style}">{text}</mstts:express-as>'
+
+    @staticmethod
+    def role_text(role, text):
+        return f'<mstts:express-as role="{role}">{text}</mstts:express-as>'
+
+    @staticmethod
+    def style_text(style, text):
+        return f'<mstts:express-as style="{style}">{text}</mstts:express-as>'
 
 if __name__ == "__main__":
     azure_tts = AzureTTS("azure_tts")
+    text = """
+    女儿看见父亲走了进来，问道：
+        <mstts:express-as role="YoungAdultFemale" style="calm">
+            “您来的挺快的，怎么过来的？”
+        </mstts:express-as>
+        父亲放下手提包，说：
+        <mstts:express-as role="OlderAdultMale" style="calm">
+            “刚打车过来的，路上还挺顺畅。”
+        </mstts:express-as>
+    """
+    path = WORKSPACE_ROOT / "tts"
+    path.mkdir(exist_ok=True, parents=True)
+    filename = path / "output.wav"
     azure_tts.synthesize_speech(
         "zh-CN",
         "zh-CN-YunxiNeural",
-        "Boy",
-        "你好，我是卡卡",
-        "output.wav")
+        text=AzureTTS.role_style_text(role="Boy", style="affectionate", text="你好，我是卡卡"),
+        output_file=str(filename)
+    )
diff --git a/tests/metagpt/actions/test_azure_tts.py b/tests/metagpt/actions/test_azure_tts.py
index b5a333af2..2145f7133 100644
--- a/tests/metagpt/actions/test_azure_tts.py
+++ b/tests/metagpt/actions/test_azure_tts.py
@@ -4,18 +4,38 @@
 @Time    : 2023/7/1 22:50
 @Author  : alexanderwu
 @File    : test_azure_tts.py
+@Modified By: mashenquan, 2023-8-9, add more text formatting options
 """
 from metagpt.actions.azure_tts import AzureTTS
+from metagpt.const import WORKSPACE_ROOT
 
 
 def test_azure_tts():
     azure_tts = AzureTTS("azure_tts")
-    azure_tts.synthesize_speech(
+    text = """
+        女儿看见父亲走了进来，问道：
+            <mstts:express-as role="YoungAdultFemale" style="calm">
+                “您来的挺快的，怎么过来的？”
+            </mstts:express-as>
+            父亲放下手提包，说：
+            <mstts:express-as role="OlderAdultMale" style="calm">
+                “Writing a binary file in Python is similar to writing a regular text file, but you'll work with bytes instead of strings.”
+            </mstts:express-as>
+        """
+    path = WORKSPACE_ROOT / "tts"
+    path.mkdir(exist_ok=True, parents=True)
+    filename = path / "girl.wav"
+    result = azure_tts.synthesize_speech(
         "zh-CN",
-        "zh-CN-YunxiNeural",
-        "Boy",
-        "你好，我是卡卡",
-        "output.wav")
+        "zh-CN-XiaomoNeural",
+        text=text,
+        output_file=str(filename))
+
+    print(result)
 
     # 运行需要先配置 SUBSCRIPTION_KEY
     # TODO: 这里如果要检验，还要额外加上对应的asr，才能确保前后生成是接近一致的，但现在还没有
+
+
+if __name__ == '__main__':
+    test_azure_tts()