All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

• Renamed extra_headers to additional_headers in websockets.connect() to support WebSockets version 14.0, as per documentation
• Updated speechmatics-python to require Python >= 3.9, aligning with WebSockets 14.0

speechmatics-python 2.0.3 is the last version supporting Python 3.8

• Unexpected keyword argument 'extra_headers' error in websockets.connect() by updating requirements to allow versions of websockets from 10.0 up to and including 13.1

• Added internal, Speechmatics only client message: GetSpeakers, and server message: SpeakersResult
• Added internal, Speechmatics only client method: send_message

• Refactor mutable default parameters in run function

• Remove deprecated speaker_change, channel_and_speaker_change, and speaker_change_sensitivity diarization options
• Remove speaker change deprecation warning

• Speaker change deprecation warning

• Disfluency option now exposed for batch.

• Support for adding extra headers for RT websocket

• AudioEventsConfig class now defaults to empty dict instead of empty list when types not provided

• Disfluency option is now backwards compatible.

• Support for removing words tagged as disfluency.

• Support for audio_events in Batch CLI.
• Support types whitelist for audio events.

• Support for volume_threshold audio filtering in transcription config

• Add audio_events_config to BatchTranscriptionConfig

• Add audio_events_config to BatchConfig.to_config method

• Proper flag handling for Audio Events

• Support for the Audio Events feature

• Rename metrics to asr_metrics

• Fix import errors for asr_metrics module
• Misc fixes for asr_metrics module

• Add metrics toolkit for transcription and diarization

• Add support for batch auto chapters

• Improve upload speeds for files submitted with the batch client
• Retry requests in batch client on httpx.ProtocolError

• Remove generate-temp-token option from examples and examples in docs

• Add support for batch topic detection

• Add support for batch sentiment analysis
• Add support for transcribing multiple files at once (submit_jobs)

• Fix error when language provided is whitespace

• Add support for transcript summarization
• Example of using notifications

• Pass sdk information to batch and rt requests
• Add support for providing just auth_token ConnectionSettings
• Use default URLs + .toml config in python sdk

• Fixed an issue in the batch client where jobs with fetch_url were not able to be submitted
• Fixed reading translation config from config file

• TranscriptionConfig.enable_partials defaults to False

• setting TranscriptionConfig.enable_partials bool value to a string raises exception

• Support for batch and realtime urls in config .toml files

• Added support for real-time translation
• Added --enable-translation-partials to enable partials for translation only
• Added --enable-transcription-partials to enable partials for transcription only

• Updated --enable-partials to enable partials for both transcription and translation

• Add support for multiple profiles to the CLI tool

• Add support for language identification

• Fixed an issue where transcription_config was not correctly loaded from the JSON config file
• CLI transcript output now properly handles UTF-8

• printing finals in cli now correctly deletes partials for that segment

• Type annotation for BatchSpeakerDiarizationConfig.speaker_sensitivity

• Always raise an exception on transcriber error

• Fix inconsistency in docs

• Add support for translation

• Raises ConnectionClosedException rather than returning when the websocket connection closes unexpectedly

• Add sphinx-argparse to docs build pipeline to auto-document the CLI tool

• Update the docs / help texts for the CLI tool

• .toml config file support to set the auth token with CLI config set command
• CLI config unset command for removing properties from the toml file
• --generate-temp-token option to the set/unset config command and toml file
• Default URLs for self-service Batch and RT in the CLI

• Documentation for base transcription config class _TranscriptionConfig
• Human-readable error outputs in the CLI

• Improved error types in HTTP requests to capture errors more clearly
• Remove excess logging on errors and allow developer to catch errors
• Use environment variable SM_MANAGEMENT_PLATFORM_URL before defaulting to production MP API URL

• Check for error in submit job response

• Url ending in '/v2/' does not return 404 error anymore
• Perform non-blocking reads when reading chunks from a synchronous stream

• Add --config-file CLI argument to allow passing a whole TranscriptionConfig JSON file to the transcriber

• Changed github workflow trigger to released

• Add --generate-temp-token CLI argument to rt websocket setup to get temp token for rt authentication
• Add generate_temp_token optional boolean kwarg to connection settings, defaults to False
• Add new RT self-service runtime URL for eu2

• Add --print-json CLI argument to enable printing transcripts as JSON rather than text
• Add speechmatics.adapters module with support for performing JSON to text conversion
• Add support for language_pack_info in the RecognitionStarted message

• Restored postional language parameter to TranscriptionConfig.__init__

• Support for enable entities, speaker diarization sensitivity, channel diarization labels in batch

• Transformed command to follow the pattern of RT only for legacy compatibility
• Fix client crashing if 'url' parameter is omitted and now outputting informative message
• Changed diarization option <speaker_and_channel> to <channel_and_speaker_change> as that's what SaaS expects.
• Fix get-results to fetch the transcript
• Update batch delete job to return meaningful response

• Update documentation for RT speaker diarization.

• Add support for speaker diarization in RT, and support the max_speakers parameter

• Remove support for --n_best_limit parameter

• Remove unnecessary Version file use and updated documentation for batch_client

• Added support for Batch ASR client

• Add domain parameter

• Fix an issue with an unhandled task exception when using run_synchronously with a timeout.

• Remove default values from args parser for max-delay-mode and operating-point for backwards compatibility with older versions of RTC.

• Use later version of sphinx to generate docs (supports Python 3.10)
• Update Speechmatics logo
• Allow user to raise ForceEndSession from an event handler or middleware in order to forcefully end the transcription session early.

• Publish to pypi.org not test.pypi.org.

• Update helper text for enable-entities, max-delay, and max-delay-mode

• Support for choosing mode of operation for max_delay via max_delay_mode in transcription config.

• bump websockets dependency to 10.1 to get the fix for an issue it has with Python 3.10

• bump websockets dependency to 9.1

• Support for enabling inverse text normalization (ITN) entities via enable_entities in transcription config.

• operating_point CLI option validation and documentation

• operating_point CLI option and property in TranscriptionConfig

• Fix seq_no persisting across sessions

• Migrate from Travis CI to GitHub Actions

• Added authentication token support for RT-SaaS @rakeshv247.