Media Processing

Media Processing vs Deepgram

Deepgram is a speech API built for speed and voice agents, with real-time streaming and its Nova and Flux models. Media Processing is a media toolkit: transcription and subtitles next to video and image processing, in the browser and over one API. Here is where each one is the better pick.

Side by side

Media ProcessingDeepgram
PriceFrom $9/month for 1,500 bonuses; at most 30 bonuses per file$0.0043–$0.0052 per minute ($0.26–$0.31 per hour), Nova-3 batch
What you pay forProcessing time per file: 2 bonuses, +1 per extra minute, capped at 30Per minute of audio
Free to startA daily free allowance, no card$200 credit, no card
Transcript formatsTXT, SRT, VTT, ASS, JSONJSON; SRT/VTT through a separate open-source library
Captions burned into the videoYesNo
Real-time (streaming) transcriptionNoYes
Speaker labelsNoYes, included in batch
LanguagesAbout 100 (Whisper), with auto-detection50+ (Nova-3), plus a 10-language multilingual mode
Longest file3 hours of media on paid plans, 30 minutes free2 GB per file
Browser app for people who don't codeYesDeveloper playground only
Video and image tools in the same APIYesNo

Where Deepgram is the better pick

  • Real-time streaming and models built for voice agents.
  • Speaker labels included in batch transcription.
  • $200 of free credit to test with.

Where Media Processing is the better pick

  • Subtitle files straight from the API — SRT, VTT, ASS — and captions burned into the video.
  • Video and image tools and a browser studio in the same product.
  • One price per file with a hard cap: however long it runs, a file never costs more than 30 bonuses.

Prices and features checked on 2026-10-07 from each vendor's public pages. Prices change — check the sources before you decide.

Sources: deepgram.com · developers.deepgram.com · developers.deepgram.com · github.com

Same thing over the API

One header, one multipart request with the file, then poll the job or take a webhook. The same operation you just tried in the browser.

Read the API docs · Quickstart

Questions

Is Media Processing cheaper than Deepgram?

Deepgram's Nova-3 batch costs $0.0043–$0.0052 per minute of audio, about $0.26–$0.31 an hour. Media Processing charges by processing time with a cap of 30 bonuses per file — about $0.12–$0.18 at most on the monthly plans for a file of up to three hours. For long files the cap is usually cheaper; Deepgram's $200 free credit goes a long way for testing.

Does Media Processing transcribe in real time?

No — files only. For voice agents and live captions, Deepgram's streaming models are built for exactly that.

Do I get subtitle files directly?

Yes — ask the transcription operation for SRT, VTT or ASS, or burn captions into the video. Deepgram returns JSON and offers a separate open-source library to build captions from it.

How hard is it to switch?

Both are REST APIs: send a file, poll the job or take a webhook, download the result. With Media Processing the same key also handles video and image processing.