CHANGELOG.md

Changelog

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.

[5.0.0.post1] - 2026-03-16

• Gracefully handle unknown message types

[5.0.0] - 2025-08-14

• Moved channel field to root of AddTranscript message for multichannel transcription.

[4.0.9] - 2025-07-03

• Deprecating speechmatics-python in favor of speechmatics-rt for real-time transcription and 'speechmatics-batch' for batch transcription.

[4.0.8] - 2025-07-03

• Patch WebsocketClient to work with any websockets version >= 10.0

[4.0.7] - 2025-06-19

• Support requesting a temporary token (JWT) with region and client ref

[4.0.6] - 2025-06-19

• Moved channel_diarization_labels field from realtime transcription config to common class.

[4.0.5] - 2025-06-18

• Added missing flag to call_middleware for multichannel mode.

[4.0.4] - 2025-06-09

• Fixed non-multichannel sessions bugging out after adding multichannel support

[4.0.3] - 2025-06-06

• Fixed microphone transcription tests not working after adding multichan dz support

[4.0.2] - 2025-06-04

Fixed

• Fixed version number not matching release, incremented new version number for hotfix.

[4.0.1] - 2025-04-30

Added

• Support RT Multichannel and channel DZ

[4.0.0] - 2025-05-20

Changed

BREAKING CHANGE: Metrics functionality now requires explicit installation

Previously, all metrics dependencies (pyannote, pandas, jiwer, etc.) were installed by default. This change moves them to an optional '[metrics]' extra to reduce the default installation footprint.

• Move metrics dependencies to requirements-metrics.txt
• Configure extras_require in setup.py for optional installation
• Add graceful error handling in CLI when dependencies are missing

[3.0.6] - 2025-05-20

Added

• Support end-of-utterance messages (DEL-24982)

[3.0.5] - 2025-05-15

• cli: fix some config options not being set when defined in a config file: topic_detection_config and speaker_diarization_config

[3.0.4] - 2025-04-16

• Support for new parameters prefer_current_speaker and speaker_sensitivity in Speaker Diarization

[3.0.3] - 2025-03-03

Added

• Support search/replace API (DEL-24399 DEL-24766)

[3.0.2] - 2024-12-18

Added

• Introduce Mixed Error Rate to SM Metrics

[3.0.1] - 2024-12-09

Fixed

• Language code getting added to URL query parameter

[3.0.0] - 2024-11-12

Changed

• 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

[2.0.3] - 2024-11-12

speechmatics-python 2.0.3 is the last version supporting Python 3.8

Fixed

• 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

[2.0.2] - 2024-10-04

Added

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

[2.0.1] - 2024-08-28

Changed

• Refactor mutable default parameters in run function

[2.0.0] - 2024-08-01

Changed

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

[1.15.0] - 2024-07-31

Added

• Speaker change deprecation warning

[1.14.10] - 2024-06-21

Fixed

• Disfluency option now exposed for batch.

[1.14.9] - 2024-06-14

Added

• Support for adding extra headers for RT websocket

[1.14.8] - 2024-05-14

Changed

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

[1.14.7] - 2024-04-08

Fixed

• Disfluency option is now backwards compatible.

[1.14.6] - 2024-04-26

Added

• Support for removing words tagged as disfluency.

[1.14.5] - 2024-03-20

Added

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

[1.14.4] - 2024-03-04

Added

• Support for volume_threshold audio filtering in transcription config

[1.14.3] - 2024-02-29

Fixed

• Add audio_events_config to BatchTranscriptionConfig

[1.14.2] - 2024-02-28

Fixed

• Add audio_events_config to BatchConfig.to_config method

[1.14.1] - 2024-02-21

Fixed

• Proper flag handling for Audio Events

[1.14.0] - 2024-02-12

Added

• Support for the Audio Events feature

[1.13.1] - 2023-12-21

Changed

• Rename metrics to asr_metrics

Fixed

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

[1.13.0] - 2023-12-07

Added

• Add metrics toolkit for transcription and diarization

[1.12.0] - 2023-11-03

Added

• Add support for batch auto chapters

1.11.1 - 2023-10-19

Added

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

Changed

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

1.11.0 - 2023-08-25

Added

• Add support for batch topic detection

[1.10.0] - 2023-08-02

Added

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

1.9.0 - 2023-06-07

Fixed

• Fix error when language provided is whitespace

Added

• Add support for transcript summarization
• Example of using notifications

1.8.3

Added

• 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

• 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

1.8.2

Fixed

• TranscriptionConfig.enable_partials defaults to False

1.8.1

Fixed

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

Added

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

1.8.0

Added

• 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

Changed

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

Added

• Add support for multiple profiles to the CLI tool

1.7.0 - 2023-03-01

Added

• Add support for language identification

Fixed

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

1.6.4 - 2023-02-14

Fixed

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

1.6.3 - 2023-02-14

Fixed

• Type annotation for BatchSpeakerDiarizationConfig.speaker_sensitivity

1.6.2 - 2023-02-07

Changed

• Always raise an exception on transcriber error

1.6.1 - 2023-02-02

Changed

• Fix inconsistency in docs

1.6.0 - 2023-02-02

Added

• Add support for translation

Changed

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

[1.5.1] - 2023-01-17

Added

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

Changed

• Update the docs / help texts for the CLI tool

1.5.0 - 2023-01-13

Added

• .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

1.4.5 - 2023-01-03

Added

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

Updated

• 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

1.4.4 - 2022-12-06

Added

• Check for error in submit job response

Fixed

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

[1.4.3] - 2022-11-24

Added

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

[1.4.1] - 2022-10-28

Updated

• Changed github workflow trigger to released

[1.4.0] - 2022-10-27

Added

• 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

[1.3.0] - 2022-08-05

Added

• 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

[1.2.3] - 2022-07-22

Fixed

• Restored postional language parameter to TranscriptionConfig.__init__

[1.2.2] - 2022-07-20

Added

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

Changed

• 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

[1.2.1] - 2022-06-17

• Update documentation for RT speaker diarization.

[1.2.0] - 2022-06-14

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

[1.1.0] - 2022-06-13

• Remove support for --n_best_limit parameter

[1.0.6] - 2022-06-01

• Remove unnecessary Version file use and updated documentation for batch_client

[1.0.5] - 2022-05-26

• Added support for Batch ASR client

[1.0.4] - 2022-05-19

• Add domain parameter

[1.0.3] - 2022-04-22

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

[1.0.2] - 2022-04-14

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

[1.0.1] - 2022-04-13

• 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.

[1.0.0] - 2022-03-23

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

[0.0.19] - 2022-02-24

Added

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

[0.0.18] - 2022-02-10

Added

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

[0.0.17] - 2022-01-19

Added

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

[0.0.16] - 2022-01-18

Added

• bump websockets dependency to 9.1

[0.0.15] - 2021-12-23

Added

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

[0.0.14] - 2021-09-07

Added

• operating_point CLI option validation and documentation

[0.0.13] - 2021-03-29

Added

• operating_point CLI option and property in TranscriptionConfig

[0.0.12] - 2021-02-10

Fixed

• Fix seq_no persisting across sessions

[0.0.11] - 2020-10-18

Changed

• Migrate from Travis CI to GitHub Actions

[0.0.10] - 2020-10-01

Added

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