Inbox

Drop a recording. Get the transcript, the summary, and the call filed.

A small watcher runs on Dana's own computer. Every 20 seconds it looks in the Inbox folder, and at the pocket recorder when it is plugged in. Anything new is transcribed with speaker labels, summarized, and moved into its client's folder, with the open items added to a ledger. The transcription runs on her own machine, so the recordings are never uploaded to a transcription service.

Sample data. Nothing is uploaded.

Inbox

Recording	Source	Recorded	Length	State

Drop a recording herePhone calls, screen recordings, voicemails, or the pocket recorder's raw files. In this demo the five sample recordings stand in for yours. Nothing you drop is read or uploaded.

Process the inbox

The watcher does this by itself. Press the button to watch one pass, replayed from a recorded run with its real timings.

Process the inbox

Where recordings come from

Phone calls: the call recording app saves a file named after the contact.
Screen recordings: a video call captured on the laptop.
Voicemail: exported from the phone system.
Pocket recorder: a small hardware recorder that mounts as a drive. Its raw files are wrapped into standard audio by the importer, named from the recording's real start time.

Free and on your own machineThe speech model runs on the computer's graphics card. No per-minute fees, no account, and the audio never leaves the machine.

Speaker labels and voice IDTurns are labelled Speaker 1, Speaker 2. Enrolled voices are recognised, so the tool can tell an internal session from a client call.

Filed and on a ledgerEach call lands in its client's folder with the transcript, subtitles, summary and routing record. Open items go on one ledger you tick off in your notes app.

Drop a file and forgetThe watcher polls, waits for a file to stop changing, and loads the model once for the whole batch. Leave it running.