This may be something recent, not sure as I normally don’t use the RECORD on my laptop in public areas when on a conference call. But I did twice today and gave up. But I shouldn’t have.
Typical video conference, in “RECORD” mode, but when I look at the transcript screen live during the meeting, its actually transcribing the voices around me in the public area. I see the words of the lady across the way talking to her child. Did this twice today and I gave up using the App waiting to come here to post its not recording my meetings… and now the TWIST!!!
It actually was recording my meetings. When it finished processing and summarizing, the summary was my actual meeting. Nothing with the lady 2 tables overs. Yet visually during the meeting, the TRANSCRIBE screen was putting the public’s voices around me on the screen. I was literally watching people’s words typing out from the AI listening to everyone around me and none of the voices in my actual call.
I go back and look at that transcription of the finished summary, and its only my meeting people. Not the public. What the heck!?!??!
Go figure that puzzle out Team Pocket. hahaha. Best of luck!
The Live Transcript is an initial, unprocessed version, so it might have a few errors. Once the recording is complete, Pocket will use a different model to refine and clean up the transcript.
I get that. My point was it is processing/listening to 2 different things but only outputting 1 (thankfully the correct one.)
Its producing what I want, so I don’t have an issue technically. But was pointing out the App is using unnecessary resources using the microphone on the device to transcribe something its not going to include in the summary.
If you don’t care, I don’t care. Guess I thought you would care.
We definitely care, and sorry if my last reply made it seem otherwise!
You caught an awkward technical limitation with how our desktop app handles dual-channel audio:
Live: The real-time preview streams straight from your computer’s microphone to save local resources, picking up nearby room noise instead of the call audio.
Final Processing: Once uploaded, our backend processes the system audio channel and strips away the background room noise, leaving you with the clean meeting summary.
Even though the final result was correct, watching the live screen transcribe the wrong people is obviously alarming.
There we go! That’s the response that makes me happy. haha
I originally posted not for my sake but for Pocket’s. I was concerned YOUR system was wasting resources with the dual function taking place. I want your backend system running faster and faster, not slower.
Your reply though makes me feel better that its my PC doing the legwork on the “Live” on local resources. And then flushing it down the toilet once done.
I was concerned your system was getting unnecessary drag. Thanks for the 2nd reply of clarity.
