Changelog

attention

This page documents changes in the Asynchronous Speech-to-Text API only. View all changes.

2024-01-06

Added

  • Added Whisper Fusion transcriber support that has better support for rare words. See Submit Transcription Job for more details

2024-10-08

Added

  • Premium Diarization feature goes out of Beta and becomes publicly available.

2024-10-18

Added

  • Added Forced Alignment feature
  • Added Asynchronous Speech-to-Text API forced_alignment submission option.

2024-10-02

Added

  • Updated US deployment of the Asynchronous Speech-to-Text API to support new low-cost transcription using Reverb Turbo model.
  • It can be used by specifying "transcriber": "low_cost" in the request.

2024-01-04

Added

  • Updated US deployment of the Asynchronous Speech-to-Text API to support asynchronous Translation and Summarization

2023-11-16

Added

  • Updated US deployment of the Asynchronous Speech-to-Text API language submission option to support new languages: Afrikaans, Armenian, Azerbaijani, Belarusian, Bosnian, Estonian, Galician, Icelandic, Kannada, Kazakh, Macedonian, Marathi, Nepali, Serbian, Swahili, Tagalog, Thai, Ukrainian, Urdu, Vietnamese, Welsh, and multilingual English/Spanish.
  • Updated US deployment of the Asynchronous Speech-to-Text API HIPAA-supported language list to all languages: Afrikaans, Arabic, Armenian, Azerbaijani, Belarusian, Bosnian, Bulgarian, Catalan, Croatian, Czech, Danish, Dutch, English, Estonian, Farsi, Finnish, French, Galician, German, Greek, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Kannada, Kazakh, Korean, Latvian, Lithuanian, Macedonian, Malay, Mandarin, Marathi, Nepali, Norwegian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swahili, Swedish, Tagalog, Tamil, Telugu, Thai, Turkish, Ukrainian, Urdu, Vietnamese, Welsh, and multilingual English/Spanish.
  • Updated US deployment of the Asynchronous Speech-to-Text API media file duration limits. All languages except Telugu supports file duration up to 17 hours, and Telugu supports up to 6 hours.

2023-09-07

Added

  • Updated Asynchronous Speech-to-Text API diarization_type submission option.

2023-08-10

Changed

  • Updated Asynchronous Speech-to-Text API to support English US ( en-us ) and English UK ( en-gb ) language values.

2023-07-19

Added

  • Updated Asynchronous Speech-to-Text API speakers_count submission option.

2023-05-15

Added

  • Updated Asynchronous Speech-to-Text API speaker_channels_count submission option documentation to include valid languages (en, es, fr).

2023-03-22

Added

  • Added a new job submission remove_atmospherics option to the Asynchronous Speech-to-Text API. This option enables you to remove atmospherics such as <laugh>, <affirmative> etc from the ASR output.

2023-01-02

Changed

  • Changed the submission option verbatim of async transcription job. Now it can be used with both machine and human transcribers. The option default value depends on the transcriber.
Transcriber Description
machine the default is true. To turn it off false should be explicitly provided
human the default is false To turn it on true should be explicitly provided

2022-09-24

Changed

  • All Rev AI job identifiers are now 16 characters in length (increased from 12 characters previously).

2022-09-15

Removed

  • Deprecated machine_v2 as an option for transcriber . Using machine is now the recommended option. Usage of machine_v2 will silently route to machine .

2022-08-25

Added

  • Added support for asynchronous non-English Speech-to-Text transcription in the European Union deployment.

2022-08-03

Changed

  • balance_seconds response value deprecated and replaced with free_balance , purchased_balance , total_balance and invoiced_balance values instead. The balance_seconds value will continue to be included in the response but will always have a value of 0 .

2022-06-23

Added

  • Added a new job submission skip_postprocessing option to the Asynchronous Speech-to-Text API. This option enables you to skip the post-processing steps (inverse text normalization or ITN, casing and punctuation) of a transcription job.

2022-06-22

Changed

  • Human Transcription feature
    • segments_to_transcribe minimum segment length lowered from 2 minutes to 1 minute

2022-06-07

Added

  • Rev AI is HIPAA compliant now.

2022-06-02

Added

2022-05-09

Added

  • notification_config as a replacement for the deprecated callback_url to provide a webhook callback URL for a job
  • Support for authorization headers when accessing URLs for notification_config

2022-05-09

Added

  • source_config as a replacement for the deprecated media_url to a provide a source URL for a job
  • Support for authorization headers when accessing URLs for source_config

2022-03-24

Added

  • custom_vocabulary as a possible failure for a failed job

2022-01-14

Added

  • (Open Beta) machine_v2 as an option for transcriber to run our Reverb ASR model for improved Word Error Rate.

2021-12-07

Added

  • Human Transcription feature
    • (Open Beta) transcriber option to enable asynchronous transcription job submissions to be transcribed by a human.
    • (Open Beta) verbatim option to enable asynchronous transcription job submissions
    • (Open Beta) rush option to enable asynchronous transcription job submissions
    • (Open Beta) segments_to_transcribe option to enable asynchronous transcription job submissions

2021-09-08

Security

  • Ended support for TLS 1.0 and TLS 1.1
    • Ended support for TLS 1.0 and TLS 1.1 for HTTPS on domain api.rev.ai. Any clients using these versions of TLS should upgrade to TLS 1.2.
    • Transport Layer Security (TLS) 1.0 and 1.1 are security protocols for establishing encryption channels over computer networks. Rev AI has supported TLS 1.0, 1.1, 1.2 protocols since the first API release. However, due to evolving regulatory requirements as well as new security vulnerabilities in TLS 1.0, Rev AI now requires that clients use at least TLS 1.2 and recommends that customers disable TLS 1.0 and 1.1 at the operating system level where possible.

2021-09-02

Security

  • Deprecated TLS 1.0 and TLS 1.1
    • Ending support for TLS 1.0 and TLS 1.1 for HTTPS on domain api.rev.ai on 2021-09-08. Any clients using these versions of TLS should upgrade to TLS 1.2. Clients that fail to update to TLS 1.2 by the end date will not be able to interact with the API.
    • Transport Layer Security (TLS) 1.0 and 1.1 are security protocols for establishing encryption channels over computer networks. Rev AI has supported TLS 1.0, 1.1, 1.2 protocols since the first API release. However, due to evolving regulatory requirements as well as new security vulnerabilities in TLS 1.0, Rev AI recommends that customers remove TLS 1.0/1.1 dependencies in their environments and disable TLS 1.0 and 1.1 at the operating system level where possible.
  • Updated security documentation
    • Updated security documentation to reflect the supported versions of TLS are now only TLS 1.2.

2021-08-10

Added

  • New Postman collection for developers to interact with the API. See documentation for details.

2021-05-26

Added

  • (Open Beta) Support for transcription for more languages

2021-04-22

Changed

  • Max allowed stream duration increased from 2 to 3 hours

2021-02-03

Changed

  • language job option is out of Open Beta and in General Availability.

Added

  • custom_vocabularies job option support for Rev AI's non-English languages. These are French, German, Portuguese and Spanish.

2020-12-04

Changed

  • Reverted minor breaking change introduced on November 9 involving job failure types. "duration out of range" failure type has been reverted to "duration exceeded" and a new failure type of "duration too short" was introduced to cover the minimum case. See get job endpoint documentation response schema for full enum of failures.

2020-12-02

Added

  • Any submitted job options through the Asynchronous and Streaming Speech-to-Text APIs except for Custom Vocabulary phrase lists are now returned in the Transcription Job Details object. This makes submitted job options visible to users of the API through the GET Job, GET List of Jobs, and POST Job endpoints responses.

2020-11-18

Added

2020-11-09

Changed

  • Changed "duration_exceeded" job failure type to "duration_out_of_range" to account for both too short and too long durations of files.

2020-08-31

Added

  • (Open Beta) custom_vocabulary_id option to enable job submission with the id of a pre-completed custom vocabulary

2020-08-20

Added

  • Improved parameter validation for callback_url parameters

2020-06-16

Added

  • Expanded the allowable character set for custom vocabularies to include some non-alpha characters. See documentation for details.