AAERT Best Practices Guide

Published

Transcript accuracy, speaker identification, inaudible marking, and certification practice.

How to read it yourself The American Association of Electronic Reporters and Transcribers. This is a professional guide, not law, and it binds nobody who has not adopted it.
How far you have to go A professional association publishes it. It binds nobody who has not adopted it, and we do not pretend otherwise.

Graded against this, and closed

Not open

These stay published with the reason they closed. The standard did not change; we could not meet it.

Deposition and Hearing Transcripts Acceptance check 1 on this unit is 99 percent word accuracy, which is a 1.0 percent word error rate. The best measured speech recognition on multi-speaker audio as of 31 July 2026 is 13.4 percent, and a deposition is the worst case for that measurement at once: several speakers, crosstalk, legal terminology and proper nouns. Closing that gap requires human review bounded by the length of the audio, which no tooling removes. At the published price that yields roughly 75 dollars per hour of work, and the volume needed to matter would take more hours than a week contains. The check was not lowered to make the unit sellable.