Let’s say you are in a conversation with just one other person. 15 minutes later another person joins the conversation. Pocket seems to think they are the same speaker. When I change speaker, it changes everyone. So Speaker ID needs improved, but also it be nice to be able to add a new speaker not just rename.
1 Like
hey @Lance currently you can create a new speaker + reassign segments but given feedback we have gotten we are reworking the system so it feels a lot more intentional thanks for the feedback!
1 Like
That seems rather work intensive, because I imagine I would have to go through segment everytimt a new speaker comes in or changes.
It is labor-intensive. I had to do it for a meeting where the third person came in late. I finally got it done but it was very finicky.
Any improvements would be appreciated.
I know they are working on this as I had several issues with speaker identification and have been chatting with the team. They are a bit limited as to the current way AI interprets voices but they are working on a few interesting ways to possibly work around this
I actually don’t worry about it too much unless there are a lot of tasks. It’s a hard thing. That it works at all fascinates me.
2 Likes