Audio Intelligence · Chain-of-Custody

Longspeak.
Every sound, accounted for.

Forensic audio intelligence with cryptographic provenance. Longspeak fingerprints every audio artifact, extracts its forensic metadata, and produces a tamper-evident receipt — the same admissibility spine that powers Decision Receipt, extended to voice and signal.

—Service
—Live Modules
—Uptime
SHA-256Provenance
checking service…
The Stack

Eight modules, one custody chain

Live modules run today on this host. Pipeline modules are orchestrated and awaiting their worker/model. Status is reported honestly — Longspeak never fabricates model output.

Why it matters for autonomous systems: an AI action backed by an audio source is only admissible if the audio itself is provably untampered. Longspeak gives every clip a content-addressed fingerprint that any party can independently recompute — closing the evidence gap between what was heard and what was decided.
Live Console

Run it now

Drop an audio file below — nothing is uploaded to third parties; bytes are inspected in-process and discarded. WAV files yield full header + loudness forensics. Everything here is real computation on your bytes.

longspeak · audio console no file loaded
Drop audio here
or click to choose — WAV, MP3, FLAC, OGG, M4A · up to 64 MB
results will appear here
Reference

API surface

All endpoints accept raw audio bytes (Content-Type: audio/* or application/octet-stream), or JSON {base64}.

MethodEndpointReturnsStatus
GET/healthService health + module countslive
GET/api/capabilitiesModule cataloglive
POST/api/audio/inspectProvenance receipt + format + WAV headerlive
POST/api/audio/analyzeRMS/peak dBFS, silence ratio, durationlive
POST/api/audio/transcribePipeline status (ASR model pending)pending model
POST/api/audio/identify-speakerPipeline status (voiceprint pending)pending model
WS/ (Upgrade)Real-time streaming intake (acks)pipeline
 
curl -s --data-binary @clip.wav -H "Content-Type: audio/wav" \
     https://longspeak.summitcognitive.ai/api/audio/inspect | jq