Flush transcription buffer on mute #557
Conversation
|
This PR includes the previous one? Let's work and merge the other one first, so you can rebase and then review this one. To minimize the number of changes in this one and not repeat my comments from the other one. |
|
|
||
| private void startBufferCheck() | ||
| { | ||
| new Thread(() -> { |
There was a problem hiding this comment.
This thread is not a daemon thread and will be preventing JVM from exiting. The code formatting is not correct.
And in general, adding a thread for every participant even if the participant will not be transcribed will increase the threads drastically and is a waste of resources.
Let's discuss this and maybe we can find a better approach.
There was a problem hiding this comment.
Maybe we can add one global thread that goes over transcriber sessions and go over participants every second, using the scheduled executors and flush when needed.
It is not time consuming task and there will very few to flush.
There was a problem hiding this comment.
I wonder though, that the buffer is not full which means that is less than 500ms of speech ... how many words is this ... I don't think that is more than one word. But yeah it may be the end of one and another one, so it will prevent a total of two words from being transcribed.
rpurdel
force-pushed
the
flush-transcription-buffer-on-mute
branch
from
3ad71d8
to
ae318aa
Compare
…tes) has been received in the last second
rpurdel
force-pushed
the
flush-transcription-buffer-on-mute
branch
from
f94d89a
to
bb6dd3c
Compare
No description provided.