Video intelligence

Generate a video summary

Use FastPix to generate AI-powered summaries that produce concise abstracts and key insights from new or existing videos.

Viewers skip videos when they cannot tell what the content covers, and search engines struggle to rank pages with no text description of the media. FastPix solves this by generating AI-written summaries through its multimodal in-video AI pipeline, producing concise abstracts that improve discoverability and help viewers decide whether to watch. Enable summaries at upload time with the summary object, or run them on existing assets with a PATCH to /on-demand/<mediaId>/summary. FastPix emits the video.mediaAI.summary.ready webhook when the abstract is ready.


Prerequisites

  • A FastPix account with an active workspace (Activate your account)
  • Your Access Token ID and Secret Key from the FastPix dashboard
  • A video or audio asset you can reach by HTTPS URL, or an existing mediaId in your workspace
  • A webhook endpoint if you plan to consume video.mediaAI.summary.ready events

Key concepts

  • A mediaId is the unique identifier FastPix assigns to every uploaded asset.
  • The summaryLength field controls the target word count of the generated abstract: minimum 30, maximum 250, default 100. FastPix runs summarization asynchronously and signals completion through a webhook event rather than blocking the upload response.

Generate a summary for new media

FastPix’s create media from URL and upload media from device endpoints accept a summary object that triggers AI summarization as part of the ingestion pipeline.

  1. Collect the HTTPS URL of the media file (pull-through URL method) or prepare a direct device upload.
  2. Send a POST request to the /on-demand endpoint with the summary object included.

Include these parameters in the request:

  • type: Specify whether the media is a video or audio file.
  • url: Provide the HTTPS URL of the media file for which a summary must be generated.
  • generate: Set this to "true" to enable the summary generation process.
  • summaryLength: Specify the desired summary length in words (for example, 120). This value determines how concise or detailed the abstract is.

Request body for creating new media from URL

Resquest
1{ 
2  "inputs": [ 
3    { 
4      "type": "video", 
5      "url": "https://static.fastpix.com/fp-sample-video.mp4" 
6    } 
7  ], 
8  "summary": { 
9    "generate": "true", 
10    "summaryLength": 120 
11  }, 
12  "accessPolicy": "public"
13}

Request body for creating new media by direct upload

When uploading media directly from a device or local storage, the inputs section carries the upload details and the payload looks like this:


Request
1{ 
2  "corsOrigin": "*", 
3  "pushMediaSettings": 
4  { 
5    "summary": {   
6    "generate": "true",   
7    "summaryLength": 120   
8  },   
9    "accessPolicy": "public"
10  } 
11}

Key considerations

  • Maximum summary length: 250 words.
  • Minimum summary length: 30 words.
  • If no specific summaryLength is provided, the default length is 100 words.

Generate a summary for existing media

To generate a summary on an asset already stored in your workspace, call the generate video summary endpoint.

  1. Collect the mediaId for the existing media.
  2. Send a PATCH request to the /on-demand/<mediaId>/summary endpoint, replacing <mediaId> with the ID of the existing asset.

Example request body:

1{  
2  "summary": {  
3    "generate": "true",  
4    "summaryLength": 120  
5  }  
6}

Tune summary length and detail

Use summaryLength to match the abstract to the viewer surface.

  • Short values (30–60 words) suit card UIs, grid tiles, and social previews.
  • Mid-range values (80–150 words) work for hero descriptions and email digests.
  • Long values (180–250 words) suit chapter introductions, course module overviews, and SEO-oriented description blocks.

For dense source material: lectures, panels, multi-speaker interviews, raise summaryLength so the model retains context. For short clips or marketing cutdowns, a lower value keeps the abstract on-topic.


Generate a summary from the dashboard

  1. Open the FastPix Dashboard > Products > Media > Add media.
  2. Choose Upload using video URL or Upload video or audio files (direct upload).
  1. On the Media details page, open the InVideo AI tab.
  2. Click AI Summary, then click Generate to start the analysis.
  3. FastPix analyzes your video and creates a summary based on its content.
  4. When the summary is ready, you can view it in the Summary section of the dashboard.

Retrieve the generated summary

To fetch the generated summary for a media asset, call the Get Media by ID endpoint with the mediaId.

You can also subscribe to the video.mediaAI.summary.ready webhook, which fires as soon as the abstract is ready. For the full event schema and sample payload, see video.mediaAI.summary.ready - AI Events.


Example event response

Response
1{
2  "type": "video.mediaAI.summary.ready",
3  "object": {
4    "type": "media",
5    "id": "8ee7c37d-5bb8-4964-967e-228331c5b34f"
6  },
7  "id": "7994845a-a87e-40f5-b27f-b61c28eda190",
8  "workspace": {
9    "name": "BrightCovePlayer-iOS",
10    "id": "f7a13f50-7f5c-48f4-b7b2-c901dcff61c6"
11  },
12  "status": "ready",
13  "data": {
14    "isSummaryGenerated": true,
15    "summary": "The transcript consists primarily of repetitive sounds and expressions, such as \"hmm\" and \"help,\" indicating a sense of urgency or distress. The use of \"huh?\" suggests confusion or a need for clarification. The overall tone conveys a feeling of being overwhelmed or in need of assistance, but the lack of specific context or detailed dialogue limits the understanding of the situation. The repetition of sounds may imply a struggle to articulate thoughts or emotions, highlighting a moment of vulnerability. Overall, the transcript captures a brief, chaotic moment filled with uncertainty and a call for help, leaving the audience curious about the underlying circumstances."
16  },
17  "createdAt": "2025-11-05T12:29:34.408978484Z",
18  "attempts": []
19}

Limits and considerations

  • Confirm the mediaId is correct when calling /on-demand/<mediaId>/summary, an invalid ID returns a 404.
  • For dense or multi-speaker content, raise summaryLength so the model captures enough context.
  • accessPolicy and maxResolution are optional but help manage access and delivery quality per asset.
  • Summarization runs asynchronously. Treat the POST/PATCH response as an acknowledgment, and use the video.mediaAI.summary.ready webhook or a Get Media by ID poll to retrieve the final abstract.

Use cases for video summaries

  • E-learning platforms: generate overviews of course modules.
  • Corporate training: capture key takeaways from training sessions.
  • Marketing teams: extract highlights from webinars and product demos.
  • Content review: assess the value of user-generated content at a glance.

Frequently asked questions

How long does FastPix take to generate a video summary?

Summarization runs after transcoding completes, so total time depends on source duration and encoding load. FastPix signals completion with the video.mediaAI.summary.ready webhook, subscribe to that event rather than polling.

What is the maximum value for summaryLength?

FastPix accepts a summaryLength between 30 and 250 words. Omitting the field produces a 100-word default abstract.

Can I regenerate the summary on an existing asset with a different length?

Yes. Send another PATCH to /on-demand/<mediaId>/summary with the new summaryLength. FastPix regenerates the abstract and emits a fresh video.mediaAI.summary.ready event.

Does FastPix support timestamped summaries or chapter markers?

The summary endpoint returns a single prose abstract. For timestamped segmentation, use the video chapters feature under in-video AI, which produces chapter boundaries alongside the summary.

What language does the generated summary use?

FastPix generates the summary in the dominant spoken language detected in the source audio. If your asset mixes languages, the abstract follows the language with the most speaking time.


What’s next?