feat: add structured AI-based meeting summary (#8)

.env.example

@@ -1 +1,2 @@
 DISCORD_BOT_TOKEN=your_bot_token_here
+OPENAI_API_KEY=your_openai_api_key_here

bot/main.py

@@ -11,7 +11,7 @@
 
 from meeting_state import MeetingStateManager, MeetingStatus
 from audio_recorder import AudioRecordingManager
-
+from summarizer import process_meeting_recordings, generate_structured_summary
 logging.basicConfig(
     level=logging.INFO,
     format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
@@ -322,10 +322,23 @@ async def end_meeting(interaction: discord.Interaction):
         )
 
         if recording_info:
-            response += f"👥 Recorded {recording_info['user_count']} user(s)\n"
-            response += f"📁 Saved to: `{recording_info['output_dir']}`"
-        
+            response += f"Recorded {recording_info['user_count']} user(s)\n"
+            response += f"Saved to: `{recording_info['output_dir']}`"
+
         await interaction.followup.send(response)
+
+        if recording_info and recording_info.get("user_count", 0) > 0:
+            await interaction.followup.send("⏳ Transcribing audio and generating summary...")
+            try:
+                transcript = process_meeting_recordings(recording_info["output_dir"])
+                if transcript:
+                    summary = generate_structured_summary(transcript)
+                    await interaction.followup.send(f"📝 **Meeting Summary**\n\n{summary}")
+                else:
+                    await interaction.followup.send("⚠️ No audio was captured to summarize.")
+            except Exception as e:
+                logger.error(f"Error generating summary: {e}")
+                await interaction.followup.send("⚠️ Could not generate summary. Check logs for details.")
 
         logger.info(
             f"Meeting ended: {ended_session.meeting_id} in guild {guild_id}, "

bot/summarizer.py

@@ -0,0 +1,79 @@
+import os
+import openai
+
+client = openai.OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
+
+
+def transcribe_audio(audio_path: str) -> str:
+    """Transcribes a .wav audio file to text using OpenAI Whisper."""
+    with open(audio_path, "rb") as audio_file:
+        response = client.audio.transcriptions.create(
+            model="whisper-1",
+            file=audio_file
+        )
+    return response.text
+
+
+def generate_structured_summary(transcript: str) -> str:
+    """Takes a plain-text transcript and returns a structured AI summary."""
+    if not transcript or not transcript.strip():
+        return "⚠️ No transcript content to summarize."
+
+    prompt = """
+You are an AI assistant that creates structured meeting summaries.
+
+Given the raw transcript below, produce a summary with exactly these three sections:
+
+## 🗝️ Key Topics Discussed
+- List the main subjects and themes covered.
+
+## ✅ Decisions Made
+- List any concrete decisions or agreements reached.
+
+## 📋 Action Items / Follow-ups
+- List tasks to be completed, with owner and deadline if mentioned.
+
+If a section has nothing to report, write: "None identified."
+
+Keep it concise and professional. Do not add any extra sections.
+
+---
+Transcript:
+{transcript}
+""".format(transcript=transcript.strip())
+
+    response = client.chat.completions.create(
+        model="gpt-3.5-turbo",
+        messages=[{"role": "user", "content": prompt}],
+        temperature=0.3,
+        max_tokens=800
+    )
+
+    return response.choices[0].message.content.strip()
+
+
+def process_meeting_recordings(output_dir: str) -> str:
+    """
+    Finds all .wav files in the meeting directory,
+    transcribes each one, and combines into a single transcript.
+    """
+    from pathlib import Path
+
+    wav_files = list(Path(output_dir).glob("*.wav"))
+
+    if not wav_files:
+        return ""
+
+    full_transcript = []
+
+    for wav_file in wav_files:
+        # Use filename as speaker label (e.g. user_123456_John)
+        speaker = wav_file.stem.split("_")[-1]
+        try:
+            text = transcribe_audio(str(wav_file))
+            if text.strip():
+                full_transcript.append(f"{speaker}: {text.strip()}")
+        except Exception as e:
+            full_transcript.append(f"[Could not transcribe {wav_file.name}: {e}]")
+
+    return "\n".join(full_transcript)

requirements.txt

@@ -1,3 +1,4 @@
 discord.py[voice]>=2.3.0
 PyNaCl>=1.5.0
 python-dotenv>=1.0.0
+openai>=1.0.0