0.37.0

What's Changed

Fixes

• Document Claude 1 and 2 models as deprecated.

Full Changelog: 0.36.0...0.37.0

0.36.0

What's Changed

Features

• Add possibility to access HTTP status code
  • Add status_code to all custom Error classes
  • Store the latest HTTP response in a new instance variable Client.last_response

Chore

• Change default HTTP timeout from 15.0 to 30.0
• Fix mypy errors
• Improve README

Full Changelog: 0.35.1...0.36.0

0.35.1

What's Changed

Fix

• Fix bug for pydantic v1 where environment variables were not respected
• Other minor pydantic fixes:
  • Replace deprecated gte field with gt for polling_interval
  • Correct usage of ConfigDict instead of SettingsConfigDict

Removal

• Removes LemurModel.claude_instant1_2 and LemurModel.basic as they have been sunsetted by Anthropic

Full Changelog: 0.35.0...0.35.1

0.35.0

What's Changed

Features

• Add Multichannel transcription support:
  • Add multichannel config field
  • Add multichannel response field
  • Add channel field to all responses with speaker field
  • Add audio_channels response field
  • Add unit tests
• Use pydantic v2
  • Pydantic v1 is still supported, but the minimum required version is bumped to pydantic>=1.10.17

Fix

• Fix custom_spelling bug where the custom_spelling property access incorrectly threw an error.
• Don't bundle tests on publish by @EmersonDove in #94

New Contributors

• @EmersonDove made their first contribution in #94

Full Changelog: 0.34.0...0.35.0

0.34.0

What's Changed

Features

• Add missing context parameter for LeMUR task by @ploeber in #91
• Add parameter to select audio input device when streaming from microphone by @mjshiggins in #82

Improvements

• Fix mypy typing errors by @ploeber in #91:
  • Add correct types in types.py, api.py, and lemur.py
  • Correct usage of httpx.codes.OK instead of httpx.codes.ok
  • Remove context property from LemurSource since self._context does not exist
• Remove httpx_mock.reset(False) in the unit tests (the interface changed and is no longer supported in Python >= 3.9)
• Add GitHub Action lint.yml with ruff formating and mypy checks by @ploeber in #85

New Contributors

• @mjshiggins made their first contribution in #82

Full Changelog: 0.33.0...0.34.0

0.33.0

Changes

Features

• New ALD features by @ploeber in #83:
  • Add language_confidence_threshold as a configuration parameter and as a JSON response field
  • Add language_confidence as a JSON response field

Improvements

• Remove SDK-side value validation in speech_threshold.setter to align with the language_confidence_threshold behavior. This lets the API handle possible errors.

Full Changelog: 0.32.0...0.33.0

0.32.0

What's Changed

Features

• Add new RealtimeError codes and avoid KeyError if the code is unsupported
• Add optional parameter return_failures to transcribe_group() and TranscriptGroup.transcribe(). If set to True, the function also returns a list of error messages for failed transcription jobs.

Improvements

• Raise Exception to user for get_by_id() and wait_for_completion() and no longer create a misleading invalid transcript object

Full Changelog: 0.31.0...0.32.0

0.31.0

What's Changed

Features

• Adds support to list transcripts (#57 ):
  • Adds functions Transcriber.list_transcripts() and Transcriber.list_transcripts_async()
  • Adds new types ListTranscriptParameters, PageDetails, ListTranscriptResponse, and TranscriptItem

Removals

• Removes conformer-2 from SpeechModel enum

Documentation

• Adds list_transcript() examples to README, including an example that shows how to paginate

Full Changelog: 0.30.0...0.31.0

0.30.0

What's Changed

Features

• Add new LeMUR models by @ploeber in #75:
  • Claude 3.5 Sonnet (LemurModel.claude3_5_sonnet)
  • Claude 3 Opus (LemurModel.claude3_opus)
  • Claude 3 Haiku (LemurModel.claude3_haiku)
  • Claude 3 Sonnet (LemurModel.claude3_sonnet)

Full Changelog: 0.29.0...0.30.0

0.29.0

What's Changed

Fixes

• Change audio_duration from float to int for the TranscriptResponse by @ploeber in #74

Full Changelog: 0.28.1...0.29.0