FAIRmat-NFDI / sand

85%

main: 84%

LAST BUILD ON BRANCH live-transcript-toggle

branch: live-transcript-toggle

CHANGE BRANCH

x

Reset

Sync Branches

• live-transcript-toggle
• always-regenerate
• arbitrary-start-samples
• async-extract-gui
• audio-companion-with-audio
• audio-entry-url-from-ids
• collect-inputs
• dashboard-cookie-login
• dashboard-entry-point
• deepgram-relay
• derived-entry
• derived-sheet-field-comments
• discard-recording
• entry-point
• expand-derived-entries
• experiments
• extract-action-pipeline
• extract-action-skeleton
• extraction-ignore-greetings
• extraction-runner
• fix-gemini-schemas
• generate
• hysprint
• hysprint-domain
• hysprint-experiment-info-out-of-service
• hysprint-sheet-store
• input-label
• inputs-list
• live-transcript-gui
• main
• misc-refactor
• nomad-api-generic-helpers
• nomad-api-refactor
• pretranscribed-audio
• read-asr-transcript
• readme-remove-url-table
• readme-rewrite
• record-layout
• record-note-layout
• request-plumbing
• resolve-sample-labels
• restore-nomad-sand-name
• retry-failed-upload
• sample-info-form
• samples-schema-fix
• session-split
• sheet-download
• sheet-sidecar-guard
• sheet-upload
• solute-relative-amount
• split-app-js
• status-step-trace
• template-cleanup
• ui-sections
• voice-commands
• voice-eln-transcribe
• voice-mode

Build # 35104522334

Build Type

Pull #49

github

Committed by yaruwangway

Commit Message

GUI toggle: choose per recording whether the live transcript is stored

A checkbox next to the recording controls (visible only when the relay
is configured): checked = the live text is sent with the audio and the
AudioInput is created pre-transcribed (whisper skipped); unchecked (the
default) = display-only, whisper transcribes. The store_live_transcript
config now only sets the toggle's default, served by the new /ui-config
endpoint; the user's last choice is remembered in the browser. The
/audio endpoint stores the transcript when the client sends it - the
field's presence is the decision.

Pull Request Pull Request #49: Per-recording toggle: store the live transcript or use whisper

Coverage Stats

1068 of 1262 relevant lines covered (84.63%)

0.85 hits per line