Hello,
We are currently using Zoom Contact Center and encountering an issue with our call recording transcripts. We would greatly appreciate it if anyone could provide insights on the correct settings or a potential workaround.
■ Issue Description
Even though the calls are conducted entirely in Japanese, the post-call recording transcripts (retrieved via the API’s transcript_url / WEBVTT, as well as the recording transcripts in “Customer Experience Analytics”) are being outputted in English. The system seems to be phonetically spelling out Japanese pronunciations into English words.
- Examples: “〜ます” (masu) → “masse” / “ka masse”, “お願い” (onegai) → “Oniga”, Customer speech → “Either of us?”, “Hi”, etc.
Upon checking our last 25 recordings, only one was correctly transcribed in Japanese.
■ Key Clue (Comparison Within the Same Call)
When analyzing the exact same call, there is a stark contrast between the two transcription features:
-
Live Transcript (Real-time during the call): Accurate Japanese (even the operator’s self-introduction is recorded perfectly).
-
Recording Transcript (Post-call via
transcript_url): Corrupted into English, making the content completely unreadable.
(Example: Engagement ID - B_GYxn6RGa8g1u8v1VjDA. We have attached the raw VTT file of the recording transcript for reference.)
■ Settings Already Checked
-
Contact Center Management > Preferences > Voice Engagement:
- “Allow call recording transcription” = Enabled
-
Same section > Live Transcription and Auto-Transcription > Language:
- Set to Japanese
-
Account Settings > Recording and Transcription:
- We could not find any specific setting to define the language for Contact Center recording transcripts.
-
Data Storage Region: Japan
■ Questions
-
Where is the setting to lock/force the post-call recording transcript language to Japanese? If the system is automatically detecting the language incorrectly and defaulting to English, is there a way to prioritize Japanese?
-
If the above is not possible, is there a way to retrieve the accurate Live Transcript via the API (Server-to-Server OAuth)? Currently, the transcript-related endpoints under the Engagement API return an empty response, and the recording object only provides the corrupted post-call recording transcript.
-
If this is a known limitation or current specification, could you let us know if there are any plans for improvement, along with a recommended workaround?
■ Environment
-
Feature Used: Zoom Contact Center (Voice)
-
Region: Japan
Thank you for your time and assistance. We look forward to hearing from you.