← Back to Blog

November Updates: Speaker Labels for Everyone

Hello!

I know, I know. Speaker labels have been sitting under "What's Next" in the last few of these posts, and it took a while to get them out the door. Out of the box solutions were not good enough. You all play for hours, and most of you have more than half a dozen people at the table. Teasing those voices apart is hard, and it had to be accurate so your notes get better, not worse. This is going to greatly improve the notes for the over half of you who record in person. Well anyway, it's out!

Speaker Labels

GM Assistant can now tell who is speaking within one recording of your whole table. No separate microphones and no special setup. The lines of your transcript are labeled with the player who said them, and your notes, recaps and memorable moments credit the right people.

Here is how it goes:

  1. List who was at the table when you upload. It fills in from your last session, so most weeks you just remove whoever missed.
  2. The analysis pauses and waits for you. Open the session and you will see a few short clips for each voice we found.
  3. Name each voice. Play a clip, pick the person. It takes about a minute.
  4. The analysis continues and your notes are written with the right names.

A few things worth knowing:

  • Unknown beats a guess - When we are not sure who said a line, it is marked with a ? and not a name. A wrong name does more damage to your notes than a missing one.
  • Your voices are not kept - The recording and the voice data used to tell speakers apart are deleted once you finish naming voices or skip the step. The Privacy Policy has the details.
  • You can skip it - Per session, or as your default in the analysis options.

Folks uploading multi-track audio have been telling us for months that their analyses make far fewer mistakes with speaker labels. That is now available to every table with one microphone. The Speaker Labels guide walks through it, with examples you can click through for the messier cases.

A big thank you to our Alpha testers, who put up with a lot of rough versions of this and told us exactly what was wrong with each one.

GMA Updates

  • Spelling and accuracy - Names and unusual words are spelled correctly more often, and the notes are more accurate overall. Your ratings and comments on individual analyses keep steering this.
  • Transcripts read better - A line is now split where the speaker changes, so a question and its answer no longer end up on the same line. We also stopped dropping some lines that only looked like repeats.
  • Smart Edit - Edits are more complete, and Smart Edit now tells you when a change could not be saved.
  • Recording consent - Before an analysis starts, everyone now sees one checkbox confirming the people in the recording agreed to it. We updated the Privacy Policy and Terms alongside it. We care a lot about your privacy, your players' rights, and doing this by the book. We are not flying by the seat of our pants over here.
  • While you wait - The session page now shows tips while an analysis is running.
  • Shared moments - New backdrop style options when you share a memorable moment.
  • Show password - A show/hide toggle on the sign-in and sign-up forms.
  • Languages - Cantonese, Mandarin, Japanese, Korean and Thai are no longer supported.
  • Small fixes - Some m4a files whose length could not be read now upload fine (we saw this with recordings made in Voice Memos on a Mac), bad audio files fail with a clear message, and your phone screen more reliably stays awake during an upload.

What's Next?

Under wraps - There is a lot coming, and we are keeping it to ourselves for now. The goal has not changed: stay easy to use, and be the most detailed and accurate option for session notes.

Tell us what you think - Use the feedback button in the app or email us. I read all of it, and a lot of what shipped this month started there. Worried someone already suggested the feature you want? Tell us anyway. We prefer not to share what others have asked for, so that we get your real thoughts and not everyone hopping on a feature bandwagon because it sounds like a good idea.

I'll be wandering around at PAX Unplugged in Philadelphia, lemme know if you're around.

-Alex

Founder, GM Assistant