Koko210/miku-discord
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
362108f4b0ed50092a2391c0cfcc65e713c70a4c
miku-discord/readmes/VOICE_CHAT_PHASE1_TESTING.md

9.6 KiB
Raw Blame History

Voice Chat Feature - Phase 1 Testing Guide

Testing Environment Setup

✅ Bot rebuilt and restarted with new dependencies:

  • PyNaCl>=1.5.0 (Discord voice support)
  • websockets>=12.0 (TTS WebSocket client)

Test Checklist

🎯 Test 1: Voice Commands Basic Functionality

1A. Test !miku join Command

Prerequisites:

  • Be in a Discord server where Miku is present
  • Join a voice channel first

Steps:

1. Join any voice channel in the server
2. Type: !miku join

Expected Results:

  • ✅ Miku joins your voice channel
  • ✅ You receive an embed message with:
    • Title: "🎤 Joined Voice Chat!"
    • Teal color (#86CECB)
    • Instructions on how to talk to Miku
    • List of 8 blocked features
  • ✅ Bot logs show all 9 resource locks activating: 
    [INFO] Switching to AMD GPU for voice session
[INFO] Vision model blocked
[INFO] Image generation disabled
[INFO] Text channels paused
[INFO] Bipolar interactions paused
[INFO] Profile picture switching locked
[INFO] Autonomous system paused
[INFO] Scheduler paused
[INFO] Figurine notifier paused

How to Check Logs:

docker-compose logs miku-bot --tail 100 | grep -i "voice\|blocked\|paused\|locked"

1B. Test !miku voice-status Command

Steps:

1. While voice session is active, type: !miku voice-status

Expected Results:

  • ✅ Embed showing:
    • Voice channel name
    • Prompt channel name (where you typed the command)
    • List of all 8 blocked features with ❌ icons

1C. Test !miku leave Command

Steps:

1. Type: !miku leave

Expected Results:

  • ✅ Miku disconnects from voice channel
  • ✅ Confirmation message: "👋 Left voice chat and restored all features!"
  • ✅ Bot logs show all 9 resource locks releasing: 
    [INFO] Switched back to GTX GPU
[INFO] Vision model unblocked
[INFO] Image generation re-enabled
[INFO] Text channels resumed
[INFO] Bipolar interactions resumed
[INFO] Profile picture switching unlocked
[INFO] Autonomous system resumed
[INFO] Scheduler resumed
[INFO] Figurine notifier resumed

🚫 Test 2: Resource Blocking - Vision Model

Steps:

1. Type: !miku join (in a text channel)
2. Upload an image or video to the text channel
3. Wait for Miku's response

Expected Results:

  • ✅ Miku responds: "🎤 I can't look at images or videos right now, I'm talking in voice chat! Send it again after I leave the voice channel."
  • ✅ Image is NOT processed
  • ✅ No vision model loading attempts in logs

🚫 Test 3: Resource Blocking - Image Generation

Steps:

1. Type: !miku join
2. Try to request an image (e.g., "draw a picture of yourself")
3. Wait for response

Expected Results:

  • ✅ Miku responds: "🎤 I can't draw right now, I'm talking in voice chat! Ask me again after I leave the voice channel."
  • ✅ Message gets a 🎤 reaction added
  • ✅ No ComfyUI requests in logs
  • ✅ Log shows: "[INFO] 🚫 Image generation blocked - voice session active"

🚫 Test 4: Resource Blocking - Profile Picture Changes

Steps:

1. Type: !miku join
2. Try to trigger a mood change that would normally change profile picture
   (This might happen automatically or through commands)

Expected Results:

  • ✅ Profile picture does NOT change
  • ✅ Logs show: "[INFO] Profile picture switching locked"
  • ✅ Error log: "Profile picture switching locked during voice session"

🚫 Test 5: Resource Blocking - Autonomous Engine

Steps:

1. Type: !miku join
2. Wait 30+ seconds (normal autonomous action interval)
3. Check if autonomous messages appear

Expected Results:

  • ✅ NO autonomous messages appear
  • ✅ Logs show: "[DEBUG] Autonomous system paused, skipping tick"

🚫 Test 6: Resource Blocking - Bipolar Mode

Prerequisites:

  • Bipolar mode must be enabled on the server

Steps:

1. Type: !miku join
2. Try to trigger a bipolar argument (this has a 15% random chance)
3. Or check if any arguments start naturally

Expected Results:

  • ✅ NO bipolar arguments start
  • ✅ Logs show: "[DEBUG] Bipolar interactions paused, skipping argument trigger"

🚫 Test 7: Resource Blocking - Scheduled Events

This test requires waiting for a scheduled event time

Expected Behavior:

  • ✅ During voice session, scheduled events (bedtime, mood rotations) should NOT fire
  • ✅ Check logs for: "Scheduler paused" / "Scheduler resumed"

🚫 Test 8: Resource Blocking - Figurine Notifier

This test requires waiting for figurine notification time

Expected Behavior:

  • ✅ During voice session, figurine notifications should NOT send
  • ✅ Check logs for: "Figurine notifier paused" / "Figurine notifier resumed"

🔄 Test 9: Session Persistence

Steps:

1. Type: !miku join
2. Leave the server text channel
3. Go to a different text channel
4. Type: !miku voice-status

Expected Results:

  • ✅ Status correctly shows the original voice channel and prompt channel
  • ✅ Session persists across channel changes

⚠️ Test 10: Error Handling

10A. Join Without Being in Voice

Steps:

1. Don't join any voice channel
2. Type: !miku join

Expected Results:

  • ✅ Error message: "You need to be in a voice channel first!"

10B. Leave When Not in Session

Steps:

1. Make sure no voice session is active
2. Type: !miku leave

Expected Results:

  • ✅ Message: "I'm not in a voice chat right now!"

10C. Join While Already in Session

Steps:

1. Type: !miku join
2. Try: !miku join again

Expected Results:

  • ✅ Error message preventing duplicate session start

🔍 Test 11: GPU Switching

This requires checking the gpu_state.json file

Steps:

1. Check current GPU state:
   cat /home/koko210Serve/docker/miku-discord/gpu_state.json

2. Type: !miku join

3. Check GPU state again:
   cat /home/koko210Serve/docker/miku-discord/gpu_state.json

4. Type: !miku leave

5. Check GPU state one more time

Expected Results:

  • ✅ Before join: {"current": "gtx"}
  • ✅ After join: {"current": "amd"}
  • ✅ After leave: {"current": "gtx"}

Command:

watch -n 2 'cat /home/koko210Serve/docker/miku-discord/gpu_state.json'

📋 Test 12: Text Channel Pausing

Steps:

1. Type: !miku join in a text channel
2. Try sending normal messages to Miku in that channel

Expected Results:

  • ✅ Miku does NOT respond to normal text messages
  • ✅ Messages get queued in TEXT_MESSAGE_QUEUE (visible in logs)
  • ✅ Only voice commands (!miku join/leave/voice-status) work

Quick Testing Script

Run all tests at once by copying this into Discord:

# Test 1: Join
!miku join

# Test 2: Check status
!miku voice-status

# Test 3: Try to block image processing
[Upload an image here]

# Test 4: Try to block image generation
can you draw a picture of a cat?

# Test 5: Leave
!miku leave

# Test 6: Verify features restored
[Upload an image here - should work now]

Monitoring Commands

Watch Bot Logs in Real-Time

docker-compose logs -f miku-bot | grep -i "voice\|block\|pause\|lock"

Check GPU State

cat /home/koko210Serve/docker/miku-discord/gpu_state.json

Check for Errors

docker-compose logs miku-bot --tail 200 | grep -i "error\|exception"

Full Voice-Related Logs

docker-compose logs miku-bot --tail 500 | grep -E "(voice|Voice|VOICE)"

Success Criteria

Phase 1 is considered PASSING if:

✅ Core Functionality:

  • Bot successfully joins voice channel
  • Bot successfully leaves voice channel
  • Status command shows accurate information

✅ Resource Blocking (5 critical checks):

  • Vision model blocked during session
  • Image generation blocked during session
  • Profile pictures locked during session
  • Autonomous engine paused during session
  • Bipolar mode paused during session

✅ System Integrity:

  • All locks properly released after leaving
  • GPU switches to AMD during session
  • No crashes or exceptions during session lifecycle
  • Text channel responses properly paused

✅ Logging:

  • All lock/unlock actions logged
  • Clear user-facing error messages
  • Debug information available for troubleshooting

Known Limitations (Phase 1)

⚠️ Expected Behaviors (Not Bugs):

  • ❌ No actual audio playback yet (Phase 2)
  • ❌ No LLM streaming to TTS yet (Phase 3)
  • ❌ Miku doesn't actually respond in voice (Phase 3)
  • ⚠️ Text input still from text channel (temporary - STT in Phase 4)

These are intentional Phase 1 limitations and will be addressed in future phases.

Troubleshooting

Issue: Bot doesn't join voice channel

Check:

# Verify PyNaCl is installed
docker-compose exec miku-bot python -c "import nacl; print('PyNaCl OK')"

# Check permissions
# Bot needs "Connect" and "Speak" permissions in Discord

Issue: Resource blocks not working

Check:

# Verify imports succeeded
docker-compose logs miku-bot | grep -i "import\|module"

# Check if voice_manager loaded
docker-compose logs miku-bot | grep "voice_manager"

Issue: Commands not recognized

Check:

# Verify command_router.py imports voice commands
docker-compose logs miku-bot | grep "voice command"

Next Steps After Phase 1 Testing

Once all tests pass, proceed to:

📋 Phase 2: Audio Streaming Implementation

  • Implement MikuVoiceSource class (discord.AudioSource)
  • Create WebSocket → Discord audio bridge
  • Handle audio format conversion
  • Test actual voice playback

See: VOICE_CHAT_IMPLEMENTATION_PLAN.md for Phase 2 details

Reference in New Issue View Git Blame Copy Permalink