feat(backend,frontend,mcp): any voice or round video can be transcribed on request, media is found by its bytes

backend/migrations/versions/b7d2e9f4a1c6_media_storage_key_index.py

@@ -0,0 +1,27 @@
+"""media storage_key index for content lookup
+
+Revision ID: b7d2e9f4a1c6
+Revises: f1c4a8e2d6b3
+Create Date: 2026-10-28 01:00:00.000000
+
+"""
+
+from collections.abc import Sequence
+
+from alembic import op
+
+revision: str = "b7d2e9f4a1c6"
+down_revision: str | None = "f1c4a8e2d6b3"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    op.execute(
+        "CREATE INDEX ix_media_storage_key ON media (storage_key) "
+        "WHERE storage_key IS NOT NULL"
+    )
+
+
+def downgrade() -> None:
+    op.execute("DROP INDEX IF EXISTS ix_media_storage_key")