Keep the original.
Capture source metadata early. Track audio, video, supplied subtitles and optimized derivatives without losing their time or identity.
Media and metadatainsonic · v0.0.0
Voice archives, speaker models, and connected knowledge.
Organize audio and video, preserve source metadata, and connect speakers to what they said. Configure local or hosted processing through a CLI-first core and desktop client.
v0.0.0 system specification. Application downloads are not available yet.
Windows · macOS · Linux CLI + desktop
“Keep the original recording.”
Source → session.wav · segment 01“Every segment stays linked to its source.”
Source → session.wav · segment 02Follow either speaker through their segment to the original audio.
From original media to connected knowledge
Capture source metadata early. Track audio, video, supplied subtitles and optimized derivatives without losing their time or identity.
Media and metadataRoute transcription, diarization and reasoning through local models or configured providers. Cueson gives subtitles one consistent contract.
Pipelines and adaptersConnect speakers, aliases and tagged audio across the library. Elect training from a frozen corpus and retrieve model versions by speaker.
Speaker audio and modelsSearch words, people and time. Follow results back to source media, navigate timelines and save graph views with optional AI query assistance.
Queries and explorationA system you can configure
Documented JSON schemas connect the CLI, adapters and application. The desktop client wraps the shared core, and versioned help travels with it.
Explore the JSON contractsStart with the system
Read the architecture, import rules and processing contracts.
Read the architectureAll documentation versionsCLI + desktop
Release packages for macOS, Windows and Linux appear here with each release. The current v0.0.0 baseline is the system specification; installable packages are not available yet.
Desktop + CLI
Desktop + CLI
Desktop + CLI