Incremental sample indexer
SampleBrain
What it is
Semantic search across an audio sample bank: CLAP embeddings, LanceDB vector index, a watch agent reconciling the disk every six hours.
The architect's call
Validated on 654 real files: 569 distinct contents, 85 duplicates caught by fingerprint. SHA-256 chosen over BLAKE3 — slower in theory, faster here because hardware acceleration on the former outweighs the latter's algorithmic edge at this file profile. Measured trap: an emptied fragment is not a deleted fragment, and the index kept ghost entries until deletion was explicitly tombstoned.
The code
Capabilities engaged
Your data cannot leave the building?
Describe the use case, the data involved and the available hardware. I answer personally within 48 business hours.