We provide a service that provides a company intranet where you can watch internal meetings async. While I’m aware that you can get a local recording with all the audio channels separated out, we prefer to import cloud recordings (they’re more likely to be available, don’t need to worry about which computer you recorded on, if it got deleted, etc…).
One of our weaknesses is lack of good speaker diaterization. We can transcribe what is said, but it’s hard to accurately identify who is saying what when we only have a single audio channel. I know that some other services will use a “zoom bot” to workaround this limitation or rely on zoom’s provided transcripts if they become available. In our case we would prefer to be able to transcribe ourselves (per-word timestamps, custom vocab, etc).
Today I was on a call with someone using Gong and I was surprised to see there was no Gong bot in the call, but we still ended up with a transcribed recording after that had a timeline of who spoke when, along with transcripts marked with speakers. Is there any new zoom api features that would make this possible? Was curious how they were doing it… if anyone knows.
Which App Type (OAuth / Chatbot / JWT / Webhook)?
OAuth/Webhook
Thanks for reaching out about this, and good question. Have you considered using our Cloud Recording APIs and Webhooks to get audio data, in conjunction with a timeline file? The timeline file will show who was actively speaking when, which might be what you’re looking for.
We’ve had some success with the timeline file for this purpose. It’s not perfect since if there’s any overlapping noise at all in the timeline, it becomes very difficult to disambiguate the speakers but it’s certainly an improvement over what we had.
My new question is… from what I understand, you enable the timeline file by enabling the " Add a timestamp to the recording" option. This seems to create a visible timestamp counter in the video. Is it possible to enable the TIMELINE file without affecting the video like that?
You should be able to use the recording.recording_audio_transcript boolean from the Update User Settings API to enable/disable the transcription for a user:
Thanks for keeping me on track, I’m not sure where I got transcription from. Sorry about that!
To actually answer your question, no there isn’t another method that will enable the TIMELINE file. If you would like the ability to enable this feature in another manner to be considered for a future release, I recommend posting in the Feature Requests category.
