Verbit MCP. Automate captioning and transcribe media from any URL.

You've got media—a video, some audio clip—and you need it transcribed clean. Forget sitting there with captions or manual transcriptions; your AI client handles the whole damn pipeline for you using the Verbit MCP Server. This thing lets you process high-quality speech-to-text jobs instantly, taking all the headaches out of media post-production.

To get started, you'll use create_job. You just send your agent a public URL pointing to any video or audio source and tell it what language you need for the transcription. The tool takes that info and kicks off a new job, immediately spitting out a unique Job ID. You can even keep track of your projects by tagging those jobs with external IDs when you create them; that keeps everything linked back to your own internal database.

Once the job starts, you'll use get_job to check its status. Just feed it that Job ID, and your agent tells you if the process is 'In Progress' or if it’s done. This lets you monitor real-time progress without having to wait around guessing what's going on in the background.

When get_job confirms the job is finished, you call get_transcript. You pass that completed Job ID and tell the tool exactly which format you need the final text. This isn't just a simple download; it lets you specify formats like JSON, SRT, VTT, DOCX, or TXT, so you get the file ready for immediate use, whether you're dropping it into an editor or archiving it.

The process is straightforward: You send your agent the URL and language via create_job. When you need a status check, you ask about the Job ID using get_job. Once that confirms completion, you hit get_transcript to download the perfectly formatted output. This workflow eliminates the need for manual captioning entirely.

When your agent uses create_job, it's not just starting a transcription; it's initiating a structured process defined by the URL and the language specification you provide, giving you that critical Job ID right out of the gate. The ability to tag jobs with external IDs means Verbit integrates into your existing project management structure—you never lose track of which transcript belongs where.

If you run into issues or just need an update, get_job is what you use. It takes that specific Job ID and returns the current status, whether it's still cooking ('In Progress') or if it's ready to be pulled down. You don't have to wait on a timer; you just check the job status until it says green.

To finish up, get_transcript is your final stop. It accepts that Job ID and lets you select from a ton of output formats: DOCX for Word docs, SRT for subtitling, VTT for web videos, JSON for machine reading, or simple TXT files. You call this tool when the job is complete to pull down the finished content in whatever format your downstream process needs.

It's about getting you the file right now, without conversion headaches.

This server handles the entire lifecycle: From the initial media upload and language definition using create_job, through the status monitoring with get_job, right up to downloading a perfectly formatted, usable transcript via get_transcript. It's a complete system for transforming audio and video into structured text files.