Releases: HumeAI/hume-python-sdk
Releases · HumeAI/hume-python-sdk
hume 0.3.7
- The
raise_on_failed
inBatchJob
will raise an exception if aBatchJob
fails - When using
HumeStreamClient
models can be configured separately for each payload. Configuring models when initializing theStreamSocket
is still supported. - Support for
fps_pred
,prob_threshold
, andmin_face_size
when usingFaceConfig
withHumeStreamClient
hume 0.3.6
- Improved error messages in
HumeBatchClient
hume 0.3.5
- When using
HumeBatchClient
to upload local files, filenames are now correctly associated with predictions.
hume 0.3.4
- Adds
timeout
parameters to bothHumeStreamClient
andHumeBatchClient
to make request timeouts configurable. - Makes improvements to docstrings and examples in docs
hume 0.3.3
- Fixes a bug in
HumeBatchClient.submit_job()
where jobs using thefiles
parameter would be configured with the defaultLanguageConfig
even if configuration params (e.g.granularity
) were specified.
hume 0.3.2
- Submit batch jobs from local files
- Support for batch job email notifications
- Facemesh model support for
HumeBatchClient
hume 0.3.1
- Bug fixes
hume 0.3.0
- Webhook support with the
callback_url
parameter inHumeBatchClient.submit_job
- New
granularity
parameter in theProsodyConfig
- Get job predictions without downloading to a file with
BatchJob.get_predictions()
hume 0.2.0
- Error codes (cross reference with error code docs)
- Named entity recognition support for
HumeBatchClient
- Improved type hints for
StreamSocket
- Support for new streaming connection parameters:
- Resize the streaming context window with
stream_window_ms
- Reset streaming connection context with
reset_stream
- Get job details with
job_details
- Resize the streaming context window with
hume 0.1.7
- Fix for language model
sliding_window
parameter, updated togranularity
parameter - Improved error messages