feat: add migrations, test scripts, and utility tools

- Add database migrations (006-028) for face recognition, identity, file_uuid
- Add test scripts for ASR, face, search, processing
- Add portal frontend (Tauri)
- Add config, benchmark, and monitoring utilities
- Add model checkpoints and pretrained model references

test_faster_whisper.py

@@ -0,0 +1,132 @@
+#!/opt/homebrew/bin/python3.11
+"""Test faster_whisper transcription in isolation."""
+
+import sys
+import os
+import time
+import tempfile
+import subprocess
+
+# Add scripts directory to path
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+
+
+def test_faster_whisper():
+    print("Testing faster_whisper...")
+
+    # Try to import
+    try:
+        from faster_whisper import WhisperModel
+
+        print("✓ faster_whisper imported successfully")
+    except ImportError as e:
+        print(f"✗ Failed to import faster_whisper: {e}")
+        return
+
+    # Load model
+    print("Loading Whisper model (tiny, int8)...")
+    start = time.time()
+    try:
+        model = WhisperModel("tiny", device="cpu", compute_type="int8")
+        elapsed = time.time() - start
+        print(f"✓ Model loaded successfully in {elapsed:.2f}s")
+    except Exception as e:
+        print(f"✗ Model loading failed: {e}")
+        import traceback
+
+        traceback.print_exc()
+        return
+
+    # Create a test audio file (1 second of silence)
+    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as f:
+        temp_wav = f.name
+
+    try:
+        # Create silent audio using ffmpeg
+        cmd = [
+            "ffmpeg",
+            "-f",
+            "lavfi",
+            "-i",
+            "anullsrc=r=16000:cl=mono",
+            "-t",
+            "1",
+            "-acodec",
+            "pcm_s16le",
+            temp_wav,
+            "-y",
+        ]
+        result = subprocess.run(cmd, capture_output=True)
+        if result.returncode != 0:
+            print(f"✗ Failed to create test audio: {result.stderr.decode()}")
+            # Try alternative: extract a small chunk from a known video
+            print("Trying to extract 5-second chunk from test video...")
+            test_video = "/Users/accusys/test_video/20250209_212949.mp4"
+            if os.path.exists(test_video):
+                cmd = [
+                    "ffmpeg",
+                    "-i",
+                    test_video,
+                    "-t",
+                    "5",
+                    "-acodec",
+                    "pcm_s16le",
+                    "-ar",
+                    "16000",
+                    "-ac",
+                    "1",
+                    temp_wav,
+                    "-y",
+                ]
+                result = subprocess.run(cmd, capture_output=True)
+                if result.returncode != 0:
+                    print(f"✗ Failed to extract audio: {result.stderr.decode()}")
+                    os.unlink(temp_wav)
+                    return
+            else:
+                print("Test video not found, skipping transcription test")
+                os.unlink(temp_wav)
+                return
+
+        print(
+            f"✓ Created test audio file: {temp_wav} ({os.path.getsize(temp_wav)} bytes)"
+        )
+
+        # Try transcription
+        print("Testing transcription...")
+        start_trans = time.time()
+        try:
+            # Use beam_size=5 like in the ASR processor
+            segments, info = model.transcribe(temp_wav, beam_size=5)
+            elapsed_trans = time.time() - start_trans
+            print(f"✓ Transcription initiated in {elapsed_trans:.2f}s")
+
+            # Convert generator to list to actually run the transcription
+            print("Converting segments to list...")
+            segments_list = list(segments)
+            elapsed_total = time.time() - start_trans
+            print(f"✓ Transcription completed in {elapsed_total:.2f}s")
+            print(f"  Segments: {len(segments_list)}")
+            print(
+                f"  Language: {info.language}, Probability: {info.language_probability}"
+            )
+
+            for i, segment in enumerate(segments_list[:3]):  # Show first 3 segments
+                print(
+                    f"  Segment {i}: {segment.start:.2f}s - {segment.end:.2f}s: {segment.text}"
+                )
+
+        except Exception as e:
+            print(f"✗ Transcription failed: {e}")
+            import traceback
+
+            traceback.print_exc()
+
+    finally:
+        if os.path.exists(temp_wav):
+            os.unlink(temp_wav)
+            print("✓ Cleaned up temp file")
+
+
+if __name__ == "__main__":
+    test_faster_whisper()