Skip to content

Audio capture is fused to the dictation loop, and it blocks #751 and #756 #760

Description

@kacperpaczos

#751 and #756 both stop at the same place, so it is probably worth naming on its own.

Audio capture in recognition_manager.py is fused to the dictation loop: it opens the device, waits for silence on a 2 second timeout, and hands a buffer to the recogniser. That is exactly right for dictation and exactly wrong for anything else. Reading a system-audio source, or transcribing an hour of a conversation, needs an audio source that does not segment on silence and does not stop when somebody stops talking.

I am not going to propose a structure — it is your file and there are 3506 lines of it. Just noting that both of those issues run into the same thing, and that whatever unblocks one unblocks the other.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions