Remove capture page, and from the moment one start speaking, a memory in progress should appear.

[josancamon19]

Is your feature request related to a problem? Please describe.

• I want to make sure memories are not lost
• A better onboarding experience, when having your first memory automatically created.
• UI simplification, removing one whole tab, and having memories as main.

Describe the solution you'd like This requires the backend to determine when a memory is created.

• How to determine new memory id.
• So the websocket once it receives transcript, it should schedule a 2 minutes event that finishes the conversation.
• Once the conversation is finished, the memory object should be sent back in the websocket to the app.
• The mobile UI should handle this by having in progress memories + removing capture/page.dart.
• Websocket reconnection has to handle "memory in progress" scenarios.

[josancamon19]

The idea here @beastoin is to modify the websocket behavior a little bit.

1. The backend should have the 2 minutes timer determining when to create a memory, not the app.
2. The app would handle receive events from the websocket and act accordingly, this will be either new segments of a memory, a new memory was created, etc.
3. Post processing should be handled on the backend side, this means we should be able to keep a temporal version of the file while a memory is in progress.
4. capture/page.dart should be removed, instead we should have only memories page, and when a new memory starts, it should appear there, on top as in-progress. -> on memory detail opened, it should show summary in progress, but show the current transcript as it goes.

[beastoin]

interesting! lemme 🏄‍♀️

[josancamon19]

More context.

[beastoin]

PR: https://github.com/BasedHardware/Omi/pull/684