FAQ¶
Last updated: 2026-09-15
Requests, Limits, and Quota¶
What audio formats are supported?¶
MP3 and WAV only; anything else returns 415 Unsupported Media Type. See Supported Audio and Limits.
What's the maximum file size?¶
The limit is ≤ 10 MB per upload; larger files return 413 Payload Too Large.
How do I check my remaining quota?¶
You may check your remaining quota through the developer portal at makimoto.ai. GET /v1/transcriptions/usage reports the same figures, but it's authenticated with your dashboard sign-in session, not an API key, so an API-key client can't call it directly; track quota from the error.details on a 429 QUOTA_EXCEEDED response instead. See the note on GET /v1/transcriptions/usage for detail.
The free allowance is 1,000 minutes of audio per month.
Do failed jobs count against my quota?¶
No. Per the quota constraint in Supported Audio and Limits, failed jobs are not counted.
Errors¶
Why am I getting a 401?¶
The API key is missing, invalid, expired, or revoked. See Key Lifetime and Rotation for how keys expire and how to issue a fresh one.
Why am I getting a 429?¶
The account's monthly transcription-minute quota has been exceeded (QUOTA_EXCEEDED). The error's details carry limit_minutes, used_minutes, and file_minutes, so you don't need a separate call to see how much is remaining. GET /v1/transcriptions/usage reports the same figures, but only to a dashboard sign-in session, not an API key; see How do I check my remaining quota? above.
Why am I getting a 415?¶
The uploaded file isn't MP3 or WAV. See Supported Audio and Limits.
Transcripts and Diarisation¶
How does speaker diarisation work?¶
Diarisation is currently stereo-only: the pipeline splits the two audio channels (left/right) and diarises by channel, rather than separating speakers within a single mixed signal.
See Diarisation for detail, including how each channel maps to a speaker index.
Can I delete a transcription?¶
Yes, DELETE /v1/transcriptions/{job_id} removes the source audio from storage immediately and flags the job for scheduled deletion per the retention policy. See its reference.
Summaries and Tags¶
Can I summarise or tag a transcription?¶
Yes. Once a transcription job has succeeded, call POST /v1/summarize or POST /v1/tag with its job_id. Each returns a new job id; poll that one, not the transcription's.
Can I summarise or tag plain text without transcribing audio first?¶
Yes. Pass transcript_text instead of transcription_job_id in the request body, and skip transcription entirely. Provide exactly one of the two; giving both, or neither, returns 400.
Why did my summarise or tag request return a 409?¶
The source transcription hasn't reached succeeded yet (TRANSCRIPTION_NOT_READY). Poll the transcription until it finishes, then retry. This only applies to the transcription_job_id path; a transcript_text request has no transcription to wait on.
Listing Jobs¶
How do I page through a large number of jobs?¶
GET /v1/transcriptions returns 10 jobs per page by default (up to 100 with ?limit=). Pass the next_cursor from one response back as ?cursor= to fetch the next page; next_cursor is null once you've reached the last one. See GET /v1/transcriptions for the full set of query parameters, including the status, type, language, created_after, and job_id filters.
Access and Data¶
Where is my data hosted?¶
All data is hosted in Singapore. See Data Residency for how this supports Singapore's PDPA and MAS guidelines.
Does Kawa support real-time transcription?¶
Not yet. The current API is post-conversation only (recorded calls, voicemail, archived interactions).
A real-time API is planned; see How It Works and the roadmap.