Q: Hi, please, I did see the previous message in answer of a comment: "Lastly, can the candidate provide a ...
recorded video that it can be uploaded for analysis?
-> Yes, we have upload capability. But we do need to do speaker diarization so there is some performance loss compared to the scribe (which can utilize the speaker intervals). We also don't have the names of the speakers so they will be marked by an id number. We'll give the option to change the name very soon."
Please, about the upload process, I dod not use the platform integrated with Spiky, how is the loss of quality from analytics?
Thank you
Iwan
Ayda_Spiky.ai
May 14, 2024A: Hi Iwan, as we explained above, the upload file does not contain the speaker information so we have to do the speaker diarization ourselves algorithmically, therefore, there is a drop in performance compared to the bot. We are just noting this down as part of the pipeline, not saying that the performance is low.
IwanGlainPLUS
Verified purchaser
Hi Ayda,
Ayda, thank you, I want to make sure as the only software I use is Sessions.us, the result will be same after all optimizations?
Ayda, thank you, sorry for questioning that
Thank you
Iwan
Hi Iwan,
Yes, the rest of the pipeline is the same.
Thanks!
IwanGlainPLUSVerified purchaser
Hi, Ayda,
Ayda, I mean, the results from imported meeting will have the same analytic efficiency as if Spiky did participate the meeting?
Thank you, Ayda
Iwan
Hi Iwan,
As I said, the voices need to be separated when it is an upload. But other than that the algorithms are the same for the whole pipeline so the rest will have the same algorithmic performance.
IwanGlainPLUSVerified purchaser
Hi, Ayda,
Ayda, thank you
Ayda, the voice are separated, but who have to separate and identify the voices, it is ourselves by going through all recordings ourselves, or it has a rendering process and just have to add the name through the identified voices?
Thank you,
Iwan
Hi Iwan,
We do the separation. But there will be no names from the recordings (naturally). The part we said we would add is the UI change to add names to identified speakers.