QUICK_TEST.md

🚀 Quick Test Guide - Visual Walkthrough

Watch the Process (10 frames, 20 seconds)

What You Just Saw

Frame-by-Frame Breakdown:

1. Frame 1: Open Google Cloud Shell

   • Free tier, no installation needed
   • https://shell.cloud.google.com/

2. Frame 2: Download test script

   curl -O https://huggingface.co/marcosremar2/ensemble-tts-annotation/raw/main/scripts/test/launch_gcp_spot.sh
   

3. Frame 3: Make executable

   chmod +x launch_gcp_spot.sh
   

4. Frame 4: Run script - finds cheapest instance

   • Scans: e2-micro, e2-small, e2-medium, e2-standard-2
   • Selects: e2-medium (~$0.01/hr)

5. Frame 5: Create spot instance

   • Instance: ensemble-test-XXXXXX
   • Zone: us-central1-a
   • External IP assigned

6. Frame 6: Install dependencies

   • Updates packages
   • Installs Python, git
   • Clones repository
   • Installs torch (CPU), transformers, librosa

7. Frame 7: Run tests

   • Test imports
   • Test annotator creation
   • Test model structure

8. Frame 8: Results ✅

   • All tests PASS
   • Time: ~8 seconds
   • Cost: $0.002

9. Frame 9: Cleanup

   • Delete instance command
   • Stop all charges

10. Frame 10: Summary

    • Total cost: $0.002 (less than 1 cent!)
    • Total time: ~5 minutes
    • System validated ✅

---

📋 Quick Copy-Paste Commands

One-liner (All Steps):

curl -O https://huggingface.co/marcosremar2/ensemble-tts-annotation/raw/main/scripts/test/launch_gcp_spot.sh && chmod +x launch_gcp_spot.sh && ./launch_gcp_spot.sh

Individual Steps:

# Step 1: Download
curl -O https://huggingface.co/marcosremar2/ensemble-tts-annotation/raw/main/scripts/test/launch_gcp_spot.sh

# Step 2: Make executable
chmod +x launch_gcp_spot.sh

# Step 3: Run
./launch_gcp_spot.sh

---

💰 Cost Breakdown

Item	Cost	Notes
Cloud Shell	$0	Free tier
e2-medium spot	$0.01/hr	Only while running
Network egress	$0	Minimal data transfer
Total	~$0.002	For ~12 min test

Less than 1 penny! 💸

---

⏱️ Time Breakdown

Step	Time	Description
Setup	30s	Download + make executable
Instance creation	1 min	GCP provisions machine
Dependency install	2 min	apt-get + pip install
Clone repo	30s	Git clone from HuggingFace
Run test	10s	Actual test execution
Cleanup	30s	Delete instance
Total	~5 min	Start to finish

---

✅ What Gets Tested

Test 1: Imports

from ensemble_tts import EnsembleAnnotator  # ✓

Test 2: Annotator Creation

annotator = EnsembleAnnotator(
    mode='quick',
    device='cpu',
    enable_events=False
)  # ✓

Test 3: Model Structure

# Validates:
# - 2 models loaded (quick mode)
# - Correct weights: 0.6, 0.4
# - Names: emotion2vec, sensevoice

All tests must PASS ✅

---

🔍 Troubleshooting

"Permission denied"

chmod +x launch_gcp_spot.sh

"Project not set"

gcloud config set project YOUR_PROJECT_ID

"Quota exceeded"

Contact GCP support or use different region

Script fails

Check logs:

gcloud compute instances get-serial-port-output INSTANCE_NAME --zone=us-central1-a

---

📊 Expected Output

============================================================
TEST SUMMARY
============================================================
  imports:           ✓ PASS
  create_annotator: ✓ PASS
  model_structure:  ✓ PASS

============================================================
✓ ALL LOCAL TESTS PASSED!
============================================================

Time: 8.2 seconds
Cost: $0.002

---

🎯 Next Steps After Test Passes

1. Fine-tune models (optional):

   python scripts/training/finetune_emotion2vec.py
   

2. Run full test (with model loading):

   python scripts/test/test_quick.py
   

3. Annotate dataset:

   python scripts/ensemble/annotate_ensemble.py \
       --input marcosremar2/orpheus-tts-portuguese-dataset \
       --mode balanced \
       --device cuda
   

---

🎬 Want to Create Your Own Demo?

python scripts/demo/create_demo_gif.py

Generates:

• demos/gcp_testing_demo.gif - Animated GIF
• demos/frames/ - Individual frames (PNG)

Customize:

• Edit create_demo_gif.py
• Change colors, text, timing
• Add more frames

---

📚 More Resources

• Full Guide: GCP_TESTING_OPTIONS.md
• Testing Docs: TESTING.md
• Main README: README.md

---

Ready to test? Open Cloud Shell now! 🚀

👉 https://shell.cloud.google.com/