I quite like being *able* to transcribe TV, but it would certainly be nice to have a setting to, say, "only record these voices" (thus I could specify "only record my voice, my wife's voice" while watching TV, and still get the benefits of the Pendant, without all the noise from the TV transcription.

@Maclean D Hello, you mark this one as shipped but i got my youtube videos and stuff transcripted

I'd prefer to choose this option in my settings.

This feature is still not working.

As recently as last night, TV shows I watched were transcribed. I hope the team is still working this issue.

Make this optional please

This is complex admittedly. I do not think these recording should be disregarded, rather they should be categorized. I may ask what TV show I was watching earlier so the info is relevant and the recording is categorized as media know it is not part of my interactions with others.

seems like any audio captured can be considered valuable context. the ability for limitless (the underlying model ai stack) to identify everything as context (not just the transcript) could be super powerful. example, seems like you really enjoyed the concert last night.

now to the point, yes, I don't want tv shows (and other 'background') to clutter up the lifelog. what if it could enhance the lifelog as a filter or as additional context. Hey, you watched three hours of TV last night. How about we do a workout today? or I do want limitless to hear the podcast I listen to so I can ask questions. what was that podcast I listened to last week and they mentioned x.

this isn't going happen right now as the models aren't this mature yet, but something to keep in mind. feels like we don't want to dismiss the additional context. just need to figure out how to integrate it correctly into the lifelog. Knowing that you watched episodes 17, 18 and 19 of season 32 could come in handy someday if you're looking for a quote from Joey.

Sometimes it identifies TV/movies and says so in my daily summary. For example, it noticed that I watched "Boss Baby" with my kids last night, but it still captured all the dialogue of the whole movie. To get my summaries more sanitized I have to delete them, and sometimes it's a lot to delete and takes forever.

For TV shows, it's not always as good. For example, the other night I was watching FRIENDS with some comical situations and the summary said later that my kids discussed me hiring a str*pper (episode with Phoebe's bachelorette party), and another summary point saying that my kids called me an a$$ (again, from the same episode). Sometimes it's funny how it misunderstands, but other times, I think "I'd never share this with someone else, as it looks too d4mning..."

I’d be semi-interested in hearing how to sanely disregard tv or radio audio. After all, how would the mic be able to tell the difference between spatially present and intentional audio vs background ignorable audio?

May be a good candidate for per-source classification and then auto-drop known bad sources.