Updated on 2026-09-26 19:12 (JST)
We would like to inform you that the issue that occurred on September 25, 2026 (Friday), in which the Asynchronous HTTP speech recognition interface (v2) failed to return sentiment analysis results, has been resolved and the transition from the temporary fix to normal operation has been completed.

This issue was resolved through a temporary fix around 10:54 AM (JST) on Friday, September 25th, and we have confirmed that the sentiment analysis function is working correctly after this switchover.

■ Period of disruption
Friday, September 25, 2026, approximately 9:00 AM to 10:54 AM (JST) (Service restored through temporary measures)

■ Date and time of completion of transition to normal operation
Saturday, September 26, 2026, around 18:14 (JST)

■ Cause of the malfunction
This was because the production license for the external service used by the sentiment analysis function had become invalid, preventing the sentiment analysis process from being performed correctly.

■ Details of the response
We applied the production license, which we received again from the vendor, to the production environment and confirmed that the sentiment analysis results were returned correctly.

■ Measures to prevent recurrence
We are currently requesting the vendor to investigate the detailed circumstances that led to the license becoming invalid.
Based on the findings of this investigation, we will consider and implement measures to prevent recurrence, including strengthening monitoring of license status and the operation of the sentiment analysis function.

We sincerely apologize for any inconvenience or trouble caused to our customers.

---

Updated on 2026-09-25 11:03 (JST)
We would like to inform you of the current situation regarding the issue that occurred around 9:00 AM (JST) today with the Asynchronous HTTP speech recognition interface (v2), which was not returning sentiment analysis results.

■ Date and time of occurrence
September 25, 2026 (Friday), approximately 9:00 AM to 10:54 AM (JST) (Temporarily restored)

■ Current situation and response
We have implemented a temporary fix to the system, and the sentiment analysis function is now working correctly.

Please note that this is currently a temporary solution, and we are continuing to work towards a permanent solution and identify the cause. We will inform you again once the permanent solution is completed or if there are any further matters to announce.

We sincerely apologize again for any inconvenience and trouble caused to our customers.

---

Published on 2026-09-25 at 10:42 (JST)
Currently, there is an issue with the Asynchronous HTTP Speech Recognition API (v2) where sentiment analysis results are not being returned for requests that specify sentiment analysis.
We sincerely apologize for the great inconvenience caused to our customers.

■ Date and time of occurrence
From around 9:00 AM (JST) on Friday, September 25, 2026 to the present

■ Affected requests
Requests sent to the Asynchronous HTTP Speech Recognition interface (v2) with sentiment analysis (sentimentAnalysis=True) enabled.

■ Phenomenon
The job completes and the speech recognition results are obtained as usual, but the sentiment_analysis value (segments) is not included in the result's sentiment_analysis, and instead the following error is returned:

code: 500 / error_code: INTERNAL_SERVER_ERROR

■ Impact on billing
For requests where the sentiment analysis results in an error due to this issue, the sentiment analysis fee will not be charged (it will not be subject to billing).
*Normal charges will apply for the use of the speech recognition function if it is functioning correctly.

■ Cause and prospects for recovery
We are currently investigating the detailed cause and the expected recovery timeline.

■ Status of response
We are currently investigating the cause and working towards restoration. We will provide further updates as soon as we have progress and an estimated timeline for restoration.

We apologize for any inconvenience or trouble this may cause our customers, and we ask for your patience while we work to restore service.