---
title: "Submit Audio for Transcription"
method: POST
path: "/open/partner/ai/transcriptions/"
tags: ["Transcription API"]
---

# Submit Audio for Transcription

`POST /open/partner/ai/transcriptions/`

Submit an audio file for transcription. Pass the pre-signed download URL returned by complete-upload. Transcription runs asynchronously — use the returned `transcription_id` to poll for results.

## Headers

- `Content-Type` string, required

## Request body

- TranscriptionRequest
  - `file_url` string, required — Pre-signed download URL returned by complete-upload (M4A, MP3, WAV).
  - `params` object — Optional transcription parameters.
    - `transcribe` object
      - `language` string — BCP-47 code (en-US, zh-CN, auto); defaults to auto.
      - `model` string — Transcription model. Options include "plaud-fast-whisper", "plaud-omni-3", and "azure-fast-transcribe"
      - `detection_level` string — Language identification level ("segment" or "chapter").
    - `vad` object
      - `decode_silence` boolean — Whether to decode silent regions.
    - `diarization` object
      - `enabled` boolean — Identify and label speakers.
      - `return_embedding` boolean — Return speaker embedding vectors.

## Response `200`

Transcription task accepted

- TranscriptionCreatedResponse
  - `transcription_id` string — Identifier used to poll for results.
  - `status` 'PENDING' | 'RECEIVED' | 'STARTED' | 'PROGRESS' | 'SUCCESS' | 'FAILURE' | 'REVOKED' — `PENDING`, `RECEIVED`, `STARTED`, `PROGRESS` indicate the task is in progress — keep polling. `SUCCESS` means the data is ready. `FAILURE` and `REVOKED` are terminal failures.
  - `data` object — Empty until the transcription completes.

---

[API](https://skmtc.net/plaud/apis/authentication-tokens.md) · [All operations](https://skmtc.net/plaud/apis/authentication-tokens/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/plaud/authentication-tokens/revisions/062a14150fe6/schema)