Update app.py

app.py

@@ -11,6 +11,8 @@ from googleapiclient.discovery import build
 from youtube_transcript_api import YouTubeTranscriptApi, TranscriptsDisabled, NoTranscriptFound
 from youtube_transcript_api.formatters import TextFormatter
 from dotenv import load_dotenv
+from pytube import YouTube
+import whisper
 
 # 환경 변수 로드
 load_dotenv()
@@ -27,6 +29,7 @@ intents.guild_messages = True
 
 # 추론 API 클라이언트 설정
 hf_client = InferenceClient("CohereForAI/c4ai-command-r-plus-08-2024", token=os.getenv("HF_TOKEN"))
+whisper_client = InferenceClient("openai/whisper-large-v3", token=os.getenv("HF_TOKEN"))
 
 # YouTube API 설정
 API_KEY = os.getenv("YOUTUBE_API_KEY")
@@ -77,9 +80,15 @@ class MyClient(discord.Client):
                         replies = await generate_replies(comments, transcript)
                         await create_thread_and_send_replies(message, video_id, comments, replies, self.session)
                     else:
-                        await message.channel.send("자막을 가져올 수 없습니다. 댓글만을 기반으로 답변을 생성합니다.")
-                        replies = await generate_replies(comments, "")
-                        await create_thread_and_send_replies(message, video_id, comments, replies, self.session)
+                        await message.channel.send("자막을 가져올 수 없습니다. Whisper 모델을 사용하여 자막을 생성합니다.")
+                        transcript = await generate_whisper_transcript(video_id)
+                        if transcript:
+                            replies = await generate_replies(comments, transcript)
+                            await create_thread_and_send_replies(message, video_id, comments, replies, self.session)
+                        else:
+                            await message.channel.send("Whisper 모델로도 자막을 생성할 수 없습니다. 댓글만을 기반으로 답변을 생성합니다.")
+                            replies = await generate_replies(comments, "")
+                            await create_thread_and_send_replies(message, video_id, comments, replies, self.session)
                 else:
                     await message.channel.send("댓글을 가져올 수 없습니다.")
             else:
@@ -87,8 +96,6 @@ class MyClient(discord.Client):
         finally:
             self.is_processing = False
 
-
-
     def is_message_in_specific_channel(self, message):
         return message.channel.id == SPECIFIC_CHANNEL_ID or (
             isinstance(message.channel, discord.Thread) and message.channel.parent_id == SPECIFIC_CHANNEL_ID
@@ -113,7 +120,6 @@ def extract_video_id(url):
     logging.debug(f'추출된 비디오 ID: {video_id}')
     return video_id
 
-
 async def get_best_available_transcript(video_id, max_retries=5, delay=10):
     async def fetch_transcript(language):
         try:
@@ -157,7 +163,24 @@ async def get_best_available_transcript(video_id, max_retries=5, delay=10):
 
     return None, None
 
+async def generate_whisper_transcript(video_id):
+    try:
+        # YouTube 비디오 다운로드
+        yt = YouTube(f'https://www.youtube.com/watch?v={video_id}')
+        audio_stream = yt.streams.filter(only_audio=True).first()
+        audio_file = audio_stream.download(output_path='temp', filename=f'{video_id}.mp3')
+
+        # Whisper 모델을 사용하여 음성을 텍스트로 변환
+        with open(audio_file, "rb") as f:
+            response = whisper_client.audio_transcription(audio=f, model="openai/whisper-large-v3")
+
+        # 임시 파일 삭제
+        os.remove(audio_file)
 
+        return response['text']
+    except Exception as e:
+        logging.error(f'Whisper 자막 생성 실패: {e}')
+        return None
 
 async def get_video_comments(video_id):
     comments = []