> For the complete documentation index, see [llms.txt](https://deeptrain.gitbook.io/documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://deeptrain.gitbook.io/documentation/getting-started/transcribe-api/video-transcription.md).

# Video Transcription

{% embed url="<https://www.loom.com/share/1b9137b671d04ed78a1d880a96139dc4?t=0>" %}

## What is Video Transcription?

This is our flagship product that lets you turn ON video processing capabilities for your LLMs and AI agents. This single infra lets you use videos to train your models and even process as input.&#x20;

## Capabilities&#x20;

<details>

<summary>Audio-Visual Understanding </summary>

Understands the whole video using timestammped audio-visual transcription.&#x20;

</details>

<details>

<summary>Supports 200+ Models</summary>

Easily connects with over 200+ top language models that are available over the interent.&#x20;

</details>

<details>

<summary>Import videos from anywhere</summary>

You can import videos from anywhere - YouTube, Vimeo, self-hosted or locally hosted.

</details>

<details>

<summary>Quantative Video Compression (QVC) </summary>

QVC is an internal video compression technique that we have developed to compress videos in such a way that no data is lost at the model's side while processing it, saving over 70% cost that you might have incurred otherwise.

</details>

<details>

<summary>Supports up to 1000 seconds of processing </summary>

The v0.0.21 of Transcribe API supports over 1000 seconds of video processing in single API call.

</details>

<details>

<summary>Supports Chunking</summary>

To process videos with larger lengths, Transcribe API also supports chunking to process 1 large video through multiple parallel requests. &#x20;

</details>

<details>

<summary>Asyncronous Calling</summary>

You can do parallel and asyncronous calling to Transcribe API to process multiple videos at time.

</details>