MerkleSpeech: Public-Key Verifiable, Chunk-Localised Speech Provenance via Perceptual Fingerprints and Merkle Commitments

Researchers have unveiled 'MerkleSpeech,' a new system designed to provide public-key verifiable proof of a speech clip's origin. It uses perceptual fingerprints and Merkle tree commitments to create a cryptographic audit trail, allowing anyone to verify which parts of an audio file are original and which have been spliced or altered. The system is designed to survive common audio processing like compression, filtering, and noise addition, targeting very low false positive rates. It offers two tiers of verification: robust watermark attribution and strict cryptographic integrity.