2026-04-09 14: 22We have added a note to show the terms of service before and after the changes.

The AmiVoice API has never used registered word data for training purposes. However, the terms of service previously stated that it could be used as training data for product improvement. Now, in line with actual operation, we have explicitly stated in the terms of service and related documentation that registered word data will not be used for training.

Please see below for details.

• Terms of Use (Speech Recognition) Article 7
• Service specifications – Handling of user dictionary data

Please check the following before and after the changes to the Terms of Service.

Terms of Use (Voice Recognition) Article 7, Paragraph 2

(Before change)
Audio data uploaded by the subscriber to use this service, and text data created as a result of using this service., and word registration dataWe may use this information for the research and development of our products and services, and for quality improvement. However, we will not use it in a way that identifies the subscriber or the content of the text data. Audio data uploaded to a URL separately designated by us as a connection destination where data is not recorded, at the subscriber's option, and text data created as a result of speech recognition, will not be used for the purposes described above.

(After change)
We may use the audio data uploaded by the user to use this service, as well as the text data generated as a result of using this service, for the research and development and quality improvement of our products and services. However, we will not use the data in a way that identifies the user or the content of the text data. Audio data uploaded to a URL separately designated by us as a connection destination where data is not recorded, at the user's option, and text data generated as a result of speech recognition, will not be used for the purposes described above.