Posts
All the articles I've posted.
How to format a New York transcript: 25 lines a page
What 22 NYCRR Part 108 says about 25 numbered lines, plus a Word setup checklist for type, spacing and line numbers.
Speaker turns and Q/A: what AI still gets wrong, and how to fix it
Four places AI still mislabels speaker turns and Q/A in depositions, a quick audio check for each, and how to fix labels in the HeyDonna editor.
Gemini, Sol and Sonnet: speaker tracking is still not fixed
Speaker tracking is our users' biggest complaint. Compare the published speaker-name scores for Gemini, GPT-6.1 Sol and Claude Sonnet 5.5.
The Record Test Bench is live: first results
The Record Test Bench is live. Audio full-chain results: legal-structure F1 53.7%, speaker-name accuracy 63.9%, best verbatim WER 20.2%.