Have seen a few times where machines that are in perfect time sync with a good source the requests fail with `RequestExpired` errors.
https://docs.aws.amazon.com/transcribe/latest/dg/CommonErrors.html
While not perfect, bumping to five minutes gives more a chance that the signed requests to start streaming won't be expired.
If transcription runs slow or has issues the queue can fill up and block
all audio processing. This gives the queue a sufficent buffer and allows
it to drop audio if it eventually fills up. This was most noticable with
bad internet connections using the `awstrnascriber` where it would take
quite a while for the websocket to eventually timeout and the bin to
enter `passthrough=true`.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/688>
By using this new property, application can select exclusive caption
source. There are three source types
- Both: Inband and transcription captions are combined if exist.
This is default behavior.
- Inband: Transcription buffers will be dropped
- Transcription: Caption meta of each video buffer will be dropped
In this version, transcriberbin doesn't provide any hint
for application to help caption source decision. That can be done
by application's strategy, passthrough status or probing inband
caption meta for example.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/684>
Fix race between latency query handler and setup_transcription()
method.
Locking order of setup_transcription() is
state lock -> setup_transcription() -> settings lock
So taking state lock inside of setting lock in src_query()
can cause deadlock.
Rusoto does not implement timeouts or retries for any of its HTTP
requests. This is particularly problematic for the part upload stage of
multipart uploads, as a blip in the network could cause part uploads to
freeze for a long duration and eventually bail.
To avoid this, for part uploads, we add (a) (configurable) timeouts for
each request, and (b) retries with exponential backoff, upto a
configurable duration.
It is not clear if/how we want to do this for other types of requests.
The creation of a multipart upload should be relatively quick, but the
completion of an upload might take several minutes, so there is no
one-size-fits-all configuration, necessarily.
It would likely make more sense to implement some sensible hard-coded
defaults for these other sorts of requests.
... and replace the deprecated --all argument with --workspace.
Also use --all-targets so that code, unit tests, integration tests
and examples can all be compiled or tested in one go.
As a side effect this allows us also to handle errors more gracefully
and to reduce memory load by outputting decoded frames immediately.
Also the code was changed a bit to reduce the number of redundant mutex
lock/unlocks.
