Audio and video (A/V) overview

Audio and video (A/V) content in eCommons should be intended for access. We currently do not accept or store very large files, such as those typical of production-quality material. A/V material submitted to eCommons should be of a manageable download size, in conformance with our Collection and content policy.

You may also submit multiple versions of the same content (for example, high res and low res, or various languages). If your content already has captions and you would like them included, we accept a variety of formats including .DFXP, .srt, .vtt, and TTML. If you would like to submit captions that are outside of these formats, please contact eCommons administrators.

If your A/V files exceed our size limit, we recommend that you submit a lower resolution version. If you are looking for a program to resize videos, we find Handbrake very helpful. It's open source, available for most popular operating systems, and has robust documentation.

See Also: Performance Releases

A/V streaming

eCommons admins will create streaming versions of all A/V content and produce machine-generated captions in compliance with the eCommons Accessibility Policy (see as an example). For embedded streaming media, eCommons uses Kaltura, the same platform as Cornell's Blackboard and Canvas implementations, and CornellCast. Streaming media can be shared or embedded elsewhere, just click on the sharing icon ( ) for a static URL and embed code.


Machine-generated captions will be created for all A/V content in eCommons with spoken language, in compliance with our Accessibility Policy. Machine-generated captions are of variable quality, and often require human remediation to correct names, grammar, and specialized vocabulary. They also lack the image description that is required in captions of edited moving-image material.

eCommons administrators will provide a copy of the machine-generated captions upon request to the submitter/submitting unit of A/V material. Please provide the link to the material in any requests for captions. The machine-captioning service generates a .DFXP file, but unless this file type is specifically requested eCommons will supply a .srt file due to the simpler structure and easy editing of the .srt.

.srt files can be edited in any text editor, and they look like this:

00:00:13,290 --> 00:00:17,960
This text is the on-screen caption.

00:00:17,960 --> 00:00:22,350
All captions should be short,
but they can have two lines!

The anatomy of a caption or subtitle in a .srt file is:

number of caption
time the caption comes on (hh:mm:ss,ms) -->time caption ends (hh:mm:ss,ms)
caption text
caption text line 2 (optional)

[blank line notes the start of the next caption]

Do not:

  • Change the format of the start and end times
  • make a caption more than 2 lines
  • remove the blank line between captions

For caption best practices and tips

We recommend the following resources from the Described and Captioned Media Program (DCMP):

Human captioning service

Groups or departments with large or ongoing captioning needs may want to consider using Cornell's preferred Human Captioning Service provider.