Start with the transcript
Voxlumo produces a reviewable transcript first. Speaker labels, timestamps, summaries, subtitles and other outputs are built around that transcript rather than replacing it.
Quality & review
Accurate output still needs a clear review path.
Voxlumo produces AI-assisted output for the customer to review. Human review is not included, so the customer should know what must be checked before relying on any final deliverable.
Review names and terminology
Proper names, specialist terminology, low-volume speech, heavy overlap and unusual accents should be checked before final delivery. Add terminology or spelling guidance in the project request when available.
Review speakers on long files
Long recordings may be processed in multiple sections. Speaker labels are generated within those sections, so identity continuity across a very long file should be checked before delivery.
Treat summaries as derived output
Executive summaries, decisions, action items and answers are generated from the transcript. If the transcript is edited, the reviewed transcript should be treated as the source of truth.
Use the right delivery format
Clean TXT is best for readable text, speaker TXT for who-said-what, SRT/VTT for subtitles, and Markdown/JSON or a Meeting Pack for structured delivery.
High-stakes work needs separate safeguards
Voxlumo provides AI-assisted output only and does not include human review. Do not rely on it for legal, medical, government or regulated work unless you independently arrange the review, hosting, retention, contractual and certification safeguards required for that use case.
Simple delivery rule
Transcript first. Review second. Extras third.
For client work, review the transcript and speaker assignment before relying on subtitles, summaries or Meeting Intelligence. This keeps corrections consistent across the final deliverables.