Hey @DaveNgl thanks for the feedback i would love to give you a timeline on providing the most accurate auto speaker tagging but sadly given currently technology and research we can’t promise that and that requires you to tag users to bring the best quality matches out. We are actively doing research on the same as well and are working on better tooling to provide a better experience while tagging. As said we do still provide best in class summaries and do let us know if you have any feedback on that end thanks
3 Likes
Thanks for the update @mhirrrr! We know you have a LOT going on in the Dev side of things so thanks for always being willing to reply!
@mhirrrr When I go in to check on speakers, under Speaker Management, there are 7 of me alone. There are several other people with the same issue where there are multiples of the same speaker. Check my screenshots here: Speaker tagging - #12 by SM7
Currently, there’s no way to merge these into one single speaker.
Also, there are at least 50 Untagged Speakers. I don’t understand why there are Untagged speakers showing if there are 0 memories? For the Untagged speakers that do have memories, which speaker am I supposed to associate them with since there are currently 7 versions of me? Not to mention the other duplicated speakers?
1 Like
I don’t think it’s a tech issue that is the biggest problem. It’s a UI issue.
And honestly, it’s an issue throughout the app. It’s just complicated to figure out how to do things, and trying to do certain things one way doesn’t do what one might think it would do.
I think Pocket has some great folks working on things, and when you work on things every day, iterating and evolving, some of the UI might make sense to y’all but the options are a bit obscure for new users. I just can’t figure out how to manually tag speakers in an organized way, except to have to do it with every single conversation, and I just don’t have hours and hours of free time to do that. Pocket is supposed to be saving me time, not giving me more work.
I’m posting this not to simply complain, but to offer some perspective. Pocket seems like magic in some ways, but in other ways it’s so close to magical, but currently not useful because it doesn’t quite do what it needs to do.
Completely understand I think we have collected a lot of feedback on the same and our designer are actively working on incorporating it and making sure we provide the most seamless experience possible. Hope to have a version of this live next week!
4 Likes
Hi guys any updates on here? I’m updating individual conversations but the app is still cross-updating speaker names in other recordings
Hey @jpalusci in our list of launches for the next week we have a new app beta with revamped flows, api/webhook access, auto routing, background syncs, outlook. Speaker improvements are in the pipeline and we will keep you updated on the time line
I’m srtill having a ton of trouble with identifying speakers. Even when the ID is changed mostly correctly in the transcript …The notes still swap the name. The other person’s name is where mine should be and mine is no where to be seen.There is a reference to speaker zero that the beginning but nothing else.
I think the interface could be much simpler if we want to have speaker profiles that link across conversations. it could be pretty powerful functionality once we get it working , especially with a chatbot sitting on top of it
When editing a transcript, instead of editing a pre-assigned speaker name, or naming an unidentified speaker, can we have the option of selecting an existing speaker profile or adding a new speaker profile? This way we aren’t editing profiles in the conversations themselves (and getting this cross conversation naming changes happening).
I’m also seeing the same issue where in the summaries the speaker names aren’t showing up but sometimes Speaker0 or random speaker names plus a number. Even after re-assigning speakers and regenerating that issue persists.
Interestingly it looked like this is how the functionality was working (re-assigning an existing profile to a memory’s speakers) before the recent update, now I’m back to the previous interface where I can only rename existing speaker names which changes the master profiles and proliferates across memories
1 Like
hey @jpalusci no changes have been made on that end. If a speaker exists across speakers and a rename happens it happens across recordings. That change will come in later builds
I’m also running into an issue with speaker identification. For example, in a transcript from today with two speakers, both were tagged as the same person. When I tried to correct it, the system kept forcing both speakers to the same name. I then tried adding a new speaker, but it wouldn’t allow me to tag any sections. Now the transcript shows all speakers as untagged, and I can’t make any changes. I also tried generating a new summary, but the speakers are still untagged.
hey @Eflatt that means the segments will have to be retagged as separate users. If you can share more details on dm I can see what exactly happened