Yandex SpeechKit release notes: Speech recognition

SpeechKit provides updates by model and version.

For more information about the speech recognition methods, see About the technology.

Release as of 18/03/2026Release as of 18/03/2026

Updates to general:rc:

• Improved Russian recognition quality in certain client scenarios.
• Enhanced classification of answerphones, genders, and negativity.

Release as of 01/10/2025Release as of 01/10/2025

general:rc updates are now available in the general model.

Release as of 19/09/2025Release as of 19/09/2025

Updates to general:rc:

• Improved the recognition quality for Russian and Uzbek in certain client scenarios.
• Enhanced the answerphone classification.
• Fixed the issues with word duplication and concatenation in recognition.

Release as of 31/07/2025Release as of 31/07/2025

Added the option of querying generative text models for recognition. Learn more about this feature in Using LLMs to process recognition results.

Release as of 24/04/2025Release as of 24/04/2025

Updates to general:rc:

• Improved the recognition quality for Uzbek.
• Improved the recognition quality for the medical vocabulary in Russian.

Release as of 11/04/2025Release as of 11/04/2025

general:rc updates are now available in the general model.

Release as of 05/03/2025Release as of 05/03/2025

Improved the recognition quality for Uzbek and Russian in general:rc.

Release as of 17/12/24Release as of 17/12/24

Improved the recognition quality for Uzbek and Kazakh in general:rc.

Release as of 10/12/24Release as of 10/12/24

Updates to general:rc effective as of December 3 are now available in general.

Release as of 03/12/2024Release as of 03/12/2024

In general:rc, fixed and improved error messages when using unsupported recognition languages and audio formats.

Release as of 31/10/24Release as of 31/10/24

Improved the recognition quality for Uzbek and Turkish in general:rc.

Release as of 09/08/24Release as of 09/08/24

Updates to general:rc:

• Improved the recognition quality for Uzbek and Kazakh.
• You can now limit the recognition languages by specifying multiple values in the language_restriction field.

Release as of 26/06/24Release as of 26/06/24

The general:rc updates of June 3 are now available in general.

Improved the recognition quality for Uzbek in general:rc.

Release as of 03/06/24Release as of 03/06/24

As requested by users, general:rc recognition quality was improved for abbreviations and medical terms in Russian.

Release as of 23/04/24Release as of 23/04/24

The general:rc updates of April 9 are now available in general.

Release as of 09/04/24Release as of 09/04/24

Changed the format of classifiers in general:rc. The formal_greeting, informal_greeting, formal_farewell, informal_farewell, insult, and profanity classifiers now return results as a probability of positives. The answerphone and negative classifiers now return only the probability of positives instead of the probability of belonging to two classes.

Release as of 27/03/24Release as of 27/03/24

All general:rc updates of February 28 are now available in general.

Updates to general:rc:

• Improved the recognition quality for Uzbek.
• Improved the speaker labeling in recognition results.

Release as of 28/02/24Release as of 28/02/24

Updates to general:rc:

• Improved the recognition quality for Uzbek.
• As requested by users, improved the recognition quality for medications, car models, and tobacco products in Russian.

Release as of 27/02/24Release as of 27/02/24

All general:rc updates are now available in general.

Release as of 12/01/24Release as of 12/01/24

Added support for the speaker labeling in recognition results in general:rc.

Release as of 12/01/24Release as of 12/01/24

Improved the recognition quality for Uzbek in general:rc.

Release as of 29/12/23Release as of 29/12/23

Updates to general:rc:

1. Fixed normalization errors for certain number representations (e.g., fifteen hundred ⟶ 1500).

2. Added support for the following classifiers:

   • gender. Returns probability values for the male and female classes.
   • negative. Returns probability values for the negative and not_negative classes.
   • answerphone. Returns probability values for the answerphone and not_answerphone classes.

3. Added classifier positives for partial recognition results (ON_PARTIAL event).

Release as of 22/11/23Release as of 22/11/23

All general:rc updates are now available in general.

Release as of 10/11/23Release as of 10/11/23

Updates to general:rc:

• Updated the speech recognition model for Russian.
• As requested by users, improved the recognition quality for the names of cities in Kazakhstan.
• Enhanced the quality of speech recognition result normalization for Kazakh.
• Fixed the internal server errors when working with small audio fragments.

Release as of 06/09/23Release as of 06/09/23

Updates to general:rc:

• Fixed the issue with English words appearing when using the recognition model in Russian.
• Improved the general recognition quality for Russian.
• As requested by users, improved the recognition quality for Russian.
• Improved the general recognition quality for Uzbek.

Audio classifiers added to general:rc in the August 15, 2023 release are now available in general.

Release as of 15/08/23Release as of 15/08/23

Added support for the audio classifiers in general:rc.

Release as of 20/07/23Release as of 20/07/23

Fixed the resampling and added the new conversation metrics in general.

Release as of 07/07/23Release as of 07/07/23

Updates to general:rc:

• Fixed the two-channel audio resampling bug in the API v3.
• Added the feature of calculating conversation metrics for speech analytics. It is set up using the speech_analysis option in the StreamingOptions message.

Release as of 13/06/23Release as of 13/06/23

Fixed switching to English during Russian speech recognition in general:rc.

Release as of 07/06/23Release as of 07/06/23

Updates to general:rc:

• Improved the recognition quality for Uzbek, German, French, Dutch, Italian, Polish, and Hebrew.
• Added the number normalization for Uzbek.
• Added support for splitting text into phrases using eou_update in FullData mode.

Release as of 25/05/23Release as of 25/05/23

The May 17 release updates are now available in general.

Release as of 17/05/23Release as of 17/05/23

Updates to general:rc:

• Improved the general recognition quality for Russian.
• As requested by users, improved the recognition quality for Russian.
• Improved the recognition quality for Uzbek, German, French, Dutch, Italian, and Polish.
• Added support for the new recognition language: Hebrew (he-HE).

Release as of 14/04/23Release as of 14/04/23

Improved the recognition quality for abbreviations in Russian based on client scenarios for the general:rc model.

Release as of 16/03/23Release as of 16/03/23

The March 7 release updates are now available in general.

Release as of 07/03/23Release as of 07/03/23

For general:rc:

1. Improved the recognition quality for Uzbek.
2. Added support for the number normalization when recognizing speech in English, German, French, Italian, Spanish, and Turkish. The number normalization is also available for Kazakh in test mode.

Release as of 08/02/23Release as of 08/02/23

1. The first version of speech recognition for Uzbek is now available in the general:rc model for all API versions. Under certain acoustic conditions, Uzbek may be recognized as Kazakh. The issue will be fixed in future model releases.
2. To query the general:rc model in the API v3, you can now specify this value in the model parameter.

Release as of 20/12/22Release as of 20/12/22

For general:rc:

1. As requested by users, improved the recognition quality for medications, first and last names, and patronymics.
2. Slightly improved the recognition quality for Kazakh and Turkish.

Release as of 20/10/22Release as of 20/10/22

For general:rc:

1. Added recognition for Brazilian Portuguese; the language code is pt-BR.
2. Improved the speech recognition quality for all languages in auto recognition mode.
3. Slightly improved the recognition quality for Russian and Kazakh.

Release as of 05/10/22Release as of 05/10/22

The September 20 release updates are now available in general.

Release as of 20/09/22Release as of 20/09/22

For general:rc:

• Improved the recognition quality for Moscow districts and medications in Russian.
• Added the language classification in auto recognition mode.

The fixes are available for testing.

Release as of 29/06/22Release as of 29/06/22

1. The multi-language model is now available in general.
2. In general:rc and general, the multi-language model can accept hints on speech languages.
3. The June 7 release updates to general:rc are now available in general for Russian.

Release as of 07/06/22Release as of 07/06/22

1. Improved the punctuation and recognition of last names in general:rc.
2. The April 25 release updates are now available in general.

Release as of 25/04/22Release as of 25/04/22

Updates to general:rc:

1. Improved the recognition of such words as gasification and regasification. in Russian.
2. Added the service feedback when processing OGG-OPUS format. If a stream is not a valid audio in OPUS format, the service returns Invalid_Argument.

Release as of 19/04/22Release as of 19/04/22

1. Added the Turkish language to the multi-language speech recognition model.
2. The new API version is available for the Yandex SpeechKit streaming recognition. The old interface will still be supported; however, all new features will only be available in the API v3.

Release as of 14/03/22Release as of 14/03/22

The March 2, 2022 general:rc version is now available by the general tag.

Release as of 02/03/22Release as of 02/03/22

The general model now offers improved recognition of names, addresses, and terms as well as punctuation placement in long sentences and texts with numbers.

Further updated the general:rc model based on the user data.

Release as of 17/02/22Release as of 17/02/22

Improved the quality of the Russian-language general:rc model in the following areas:

1. Recognition of last and first names, patronymics, and addresses.
2. Recognition of customer-specific terms. The model was enhanced with the data from the user request dated February 1, 2022, and corrected based on the user data from November 9, 2021.
3. Punctuation in long sentences and texts with numbers.

Release as of 03/02/22Release as of 03/02/22

1. general:rc now supports universal mode (the "auto" language). In this mode, the model can recognize speech in one of the following languages:

   • Russian
   • Kazakh
   • English
   • German
   • French
   • Finnish
   • Swedish
   • Dutch
   • Polish
   • Portuguese
   • Italian
   • Spanish

2. New languages are also available under their own codes. The general:rc model uses an indication as a hint for language recognition. If the language is indicated explicitly, the model will use it as a hint to improve the recognition quality. Currently, it only affects the recognition quality in Russian.

Known problems: In universal mode, the recognition quality may deteriorate in the case of continuous speech without pauses.

Release as of 26/01/22Release as of 26/01/22

1. The general and general:rc recognition models for Kazakh are now available in streaming and deferred recognition modes.

2. general:rc now supports a punctuator in streaming and deferred recognition modes.

3. In deferred recognition mode, you can now work with MP3 format.