---
title: "Audio Search"
method: POST
path: "/api/v1/search/audio"
tags: ["search > search"]
---

# Audio Search

`POST /api/v1/search/audio`

Find video segments where the spoken content matches a natural-language query. Use this to find moments by what was said, for example, thank you for your time or breaking news.Returns matching audio segments with the parent video, the transcript text, and timestamps so you can jump straight to the moment.

## Headers

- `Authorization` string, required

## Request body

- SearchAudioSearchRequest — Audio search request
  - `dataset_id` string, uuid, required — The unique identifier for the dataset
  - `offset` integer — Starting index to return
  - `limit` integer — Max number of items to return
  - `metadata_filters` SearchMetadataFilters
    - `filters` SearchMetadataFiltersFiltersItems[], required
      - union
        - SearchBooleanEqualsClause
          - `key` string, required
          - `operator` '==', required
          - `value` boolean, required
        - SearchEqualsClause
          - `key` string, required
          - `operator` '==', required
          - `value` string, required
        - SearchNotEqualsClause
          - `key` string, required
          - `operator` '!=', required
          - `value` string, required
        - SearchDateTimeRangeInclusiveClause
          - `key` string, required
          - `start_utc_epoch` number, double, nullable
          - `end_utc_epoch` number, double, nullable
          - `operator` 'DateTimeRangeInclusive', required
  - `moderation_score_type` 'probability' | 'level' — Score type returned by the moderation service. - PROBABILITY: Scores in range 0-1 (uses BGE Reranker model) - LEVEL: Scores in range 0-5 (uses OpenAI model)
  - `text_query` string, required — The text query to search for
  - `skip_moderation` boolean — Skip content moderation if enabled

## Response `200`

Successful Response

- SearchAppApiRoutesSearchV1SchemaNativeVideoModelSchemaSearchResponse — Native video model search response
  - `data` SearchAppApiRoutesSearchV1SchemaNativeVideoModelSchemaSearchResponseDataItems[], required — The paginated results for native video model datasets
    - union
      - SearchImage — An image asset response
        - `created_user_id` string, required — The user that created the resource
        - `created_dt` string, date-time, required — The created datetime of the resource
        - `updated_user_id` string, required — The user that last updated the resource
        - `updated_dt` string, date-time, required — The datetime the resource was last updated
        - `coactive_image_id` string, uuid, required — The coactive image id
        - `path` string, nullable — The original path of the image
        - `metadata` object, nullable — The user provided metadata for the image
        - `moderation_score` number, double, nullable — The moderation score of the image
      - SearchAppApiRoutesSearchV1SchemaNativeVideoModelSchemaKeyframe — A keyframe asset response for native video model datasets (without shot metadata)
        - `coactive_image_id` string, uuid, required — The coactive image id
        - `video` SearchVideo, required — A video asset response
          - `created_user_id` string, required — The user that created the resource
          - `created_dt` string, date-time, required — The created datetime of the resource
          - `updated_user_id` string, required — The user that last updated the resource
          - `updated_dt` string, date-time, required — The datetime the resource was last updated
          - `path` string, nullable — The original path of the video
          - `coactive_video_id` string, uuid, required — The coactive video id
          - `metadata` object, nullable — The user provided metadata for the video
        - `audio_segment` SearchAudioSegment — Metadata about a video audio segment
          - `start_time_ms` integer, required — The start time in milliseconds of the beginning of the audio clip
          - `end_time_ms` integer, required — The end time in milliseconds of the end of the audio clip
          - `speech_to_text_transcription` string, required — The text transcription of the audio clip
        - `timestamp` integer, nullable — The timestamp in milliseconds of the keyframe in the video
        - `moderation_score` number, double, nullable — The moderation score of the image

## Other responses

- `422` — Validation Error

---

[API](https://skmtc.net/coactive/apis/api-reference.md) · [All operations](https://skmtc.net/coactive/apis/api-reference/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/coactive/api-reference/versions/c24e3835accc/schema)