delete audio

app.py

@@ -97,7 +97,7 @@ def generate_dialogue(rounds, method, role1, role2, theme, language):
 
 def main_function(password: str, theme: str, language: str, method: str, rounds: int, role1: str, role1_gender: str, role2: str, role2_gender: str):
     if password != os.environ.get("PASSWORD", ""):
-        return "错误的密码，请重新输入。", "", ""
+        return "错误的密码，请重新输入。", ""
 
     structured_dialogue = generate_dialogue(rounds, method, role1, role2, theme, language)
 
@@ -110,7 +110,7 @@ def main_function(password: str, theme: str, language: str, method: str, rounds:
         
         chatbot_dialogue.append((role1_content, role2_content))
 
-    audio_path = dialogue_to_audio(structured_dialogue, role1_gender, role2_gender)
+    # audio_path = dialogue_to_audio(structured_dialogue, role1_gender, role2_gender)
     json_output = json.dumps({"dialogue": structured_dialogue}, ensure_ascii=False, indent=4)
 
     # 儲存對話為 JSON 文件
@@ -118,73 +118,7 @@ def main_function(password: str, theme: str, language: str, method: str, rounds:
     with open(file_name, "w", encoding="utf-8") as f:
         f.write(json_output)
 
-    return chatbot_dialogue, audio_path, file_name
-
-
-def detect_language(text):
-    """
-    Simple function to detect if text is more likely English or Chinese.
-    """
-    for char in text:
-        if '\u4e00' <= char <= '\u9fff':  # range for Chinese characters
-            return "zh"
-    return "en"
-
-def dialogue_to_audio(dialogue, role1_gender, role2_gender):
-    """
-    Converts the given dialogue into an audio file using Azure's Text-to-Speech service.
-
-    Parameters:
-    - dialogue (list): List of dictionaries containing the structured dialogue.
-    - role1_gender (str): Gender of role1. Can be "male" or "female".
-    - role2_gender (str): Gender of role2. Can be "male" or "female".
-
-    Returns:
-    - str: File path to the generated audio file.
-    """
-    # Set up Azure Speech SDK
-    speech_config = SpeechConfig(subscription=AZURE_API_KEY, region=AZURE_REGION)
-
-    voices = {
-        "en": {
-            "male": "en-US-GuyNeural",
-            "female": "en-US-JessaNeural"
-        },
-        "zh": {
-            "male": "zh-TW-YunNeural",   # Taiwanese male neural voice
-            "female": "zh-TW-HsiaoYuNeural"  # Taiwanese female neural voice
-        }
-    }
-
-    ssml = '<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="zh-CN">'
-    for item in dialogue:
-        role = item['role']
-        content = item['content']
-        language = detect_language(content)
-        gender = role1_gender if role == 'role1' else role2_gender
-        voice = voices[language][gender]
-        ssml += f'<voice name="{voice}">{content}</voice>'
-    ssml += '</speak>'
-
-    # Create an audio configuration that points to an audio file.
-    audio_file = "output_audio.wav"
-    audio_output = AudioConfig(filename=audio_file)
-
-    # Create a speech synthesizer using the given settings
-    synthesizer = SpeechSynthesizer(speech_config=speech_config, audio_config=audio_output)
-
-    # Synthesize the text
-    result = synthesizer.speak_ssml(ssml)
-
-    # Check result
-    if result.reason == result.Reason.SynthesizingAudioCompleted:
-        print(f"Speech synthesized to [{result.audio_file}] for text [{ssml}]")
-    else:
-        print(f"Could not synthesize the text, reason: {result.reason}")
-
-    return audio_file
-
-
+    return chatbot_dialogue, file_name
 
 
 if __name__ == "__main__":
@@ -203,7 +137,6 @@ if __name__ == "__main__":
         ], 
         [
             gr.components.Chatbot(label="生成的對話"),
-            gr.components.Audio(type="filepath", label="對話朗讀"),
             gr.components.File(label="下載對話 JSON 文件")
         ]
     ).launch()