Search

Streamdiver automatically indexes every uploaded media asset – transcripts, speakers, keywords, entities, OCR text from images and videos, and document content. This makes your entire media library searchable by content, not just filenames.

Full-Text Search

The basic search endpoint queries across all indexed fields:

curl

curl --request GET \
  --url "https://api.streamdiver.com/v2/search?type=all&text=quarterly%20results" \
  --header "Authorization: Bearer {token}" \
  --header "Accept: application/json"

Python

results = requests.get(
    "https://api.streamdiver.com/v2/search",
    params={"type": "all", "text": "quarterly results"},
    headers={"Authorization": f"Bearer {token}"},
).json()

for item in results["data"]:
    print(item["name"], item["highlights"])

TypeScript

const results = await fetch(
  "https://api.streamdiver.com/v2/search?" +
    new URLSearchParams({ type: "all", text: "quarterly results" }),
  { headers: { Authorization: `Bearer ${token}` } }
).then((r) => r.json());

results.data.forEach((item) => console.log(item.name, item.highlights));

The type parameter controls what to search for:

Value	Description
all	Search media assets and channels
mediaasset	Search only media assets
channel	Search only channels

Targeted Field Search

By default, all indexed fields are searched. Use the fields parameter to narrow the search to specific metadata layers:

curl --request GET \
  --url "https://api.streamdiver.com/v2/search?type=mediaasset&text=budget&fields=transcript&fields=document_content" \
  --header "Authorization: Bearer {token}"

Available Search Fields

Field	What it searches
transcript	Spoken content (automatic transcription)
speakers	Recognized speaker names
entities	Extracted entities (people, places, organizations)
keywords	Automatically derived keywords
video_text	On-screen text via OCR (videos)
image_text	Text detected in images via OCR
document_content	Full text content of documents
name	Asset name
description	Asset description
tag	Assigned tags
objects	Detected objects in video/images
actions	Detected actions in video
exif_metadata	EXIF metadata from images

Filters

Combine text search with filters to narrow results:

curl --request GET \
  --url "https://api.streamdiver.com/v2/search?type=mediaasset&text=keynote&assetTypes=video&channelId={channelId}&durationTo=600" \
  --header "Authorization: Bearer {token}"

Commonly used filters:

Parameter	Description
assetTypes	Filter by video, audio, image, or document
channelId	Scope results to a specific channel
speakers	Filter by speaker name
tags	Filter by tag
durationFrom / durationTo	Duration range in seconds
after / before	Date range filters

tip

Use GET /search/ranges to retrieve the available filter ranges (max duration, file size, etc.) for your library.

Semantic Search

Beyond keyword matching, Streamdiver provides semantic search that ranks results by meaning – finding relevant content even when exact words don't match:

curl

curl --request GET \
  --url "https://api.streamdiver.com/v2/search/semantic?text=how%20to%20reduce%20costs" \
  --header "Authorization: Bearer {token}"

Python

results = requests.get(
    "https://api.streamdiver.com/v2/search/semantic",
    params={"text": "how to reduce costs"},
    headers={"Authorization": f"Bearer {token}"},
).json()

TypeScript

const results = await fetch(
  "https://api.streamdiver.com/v2/search/semantic?" +
    new URLSearchParams({ text: "how to reduce costs" }),
  { headers: { Authorization: `Bearer ${token}` } }
).then((r) => r.json());

Two semantic search modes are available:

Endpoint	Matches against
/search/semantic	Overall asset metadata (titles, descriptions, topics)
/search/textSemantic	Individual sections within assets (chapters, segments)

Pagination

All search endpoints support pagination:

/search?type=all&text=demo&perPage=10&page=2

Parameter	Description
perPage	Number of results per page
page	Page number (1-based)

Further Resources

• Interactive API Reference -- Search -- explore all search parameters
• Recommendations -- content-based recommendations
• Chat, RAG & Flows -- ask natural-language questions against your media
• Metadata -- entities, keywords, and speakers that power search

Related Use Cases

• KI-Medienanalyse -- semantic search and RAG across your entire media library
• Medienbeobachtung -- build a media monitoring platform with search and entity extraction