Action
Someone opens a door, hands over a document, points, types or walks out of frame.
Reviewed by David Dat Nguyen, CXO Last updated
Visual transcripts
What was said.
What was seen.
One transcript.
Sonix watches your video while it transcribes the speech, then writes a timestamped description of each action, scene change and piece of on-screen text beside the words.
Review body cam footage or log a day of dailies by reading. Search for a moment, then play the video from that second.
Included with every subscription plan. Start with 30 free minutes of transcription.See pricing
Every subscription plan
Timestamped
Editable
No face recognition
Four kinds of moment
Each description is a card with a kind, a start and an end time. You can change all three.
Scene change
The shot cuts, the camera turns, or the view switches from a speaker to a slide.
On-screen text
Readable words in the picture: a timestamp overlay, a slate, a sign, a slide title.
Demonstration
Someone shows how a thing works, step by step, on camera or on a shared screen.
Footage you can read
Three kinds of work where the picture matters as much as the words.
Body cam and evidence review
Speech and what the camera showed, on one timeline. Sonix describes what is visible. It never identifies a person or a face, and it never draws a conclusion.
- Read the timestamp overlay as text and jump to that second
- Find the moment a door opens or someone steps into frame
- Check a statement against what was on screen when it was made
Logging footage for the edit
A first pass through the dailies, written before you sit down to watch.
- Slates read as on-screen text, so scene and take are searchable
- Find the wide shot or the cutaway without scrubbing
- Copy a card's timecode and description into your shot log
Training and product video
Screen recordings, webinars and demos, described as they happen.
- Slide titles and button labels captured as on-screen text
- Each step of a walkthrough marked as a demonstration
- Search a webinar library for the moment a chart appeared
Video in. Every moment out.
One loop of the real workflow: upload a video, switch on the visual transcript, and read what happened beside what was said.
What the cards can do
How the cards behave in the editor, and where the descriptions travel after that.
- Beside the speech
- Cards sit in a rail next to the paragraph that was being spoken. Press play on a card and the video starts at that moment.
- Yours to correct
- Reword a description, change its kind, adjust its start and end time, or delete it. An accidental delete has an undo.
- On upload or afterwards
- Switch it on when you upload a video, or run it on a finished file from the Tools menu. If the visual pass fails, the speech transcript still opens.
- Search what was seen
- Descriptions are indexed with the transcript, so a search for a word finds it whether it was spoken or shown. Search comes with the Advanced plan and up.
Organize and search - Context for AI analysis
- AI analysis and summaries read the visual descriptions with the speech, and never quote a description as something a person said.
AI analysis - Text export and MCP
- The plain-text export carries each description with its timestamps and kind, and the MCP server returns them to your AI assistant. Subtitles, Word and PDF exports stay speech only.
MCP server
Questions teams ask about
visual transcripts
What is a visual transcript?
A visual transcript is a set of timestamped descriptions of what happens on screen in a video: actions, scene changes, on-screen text and demonstrations. Sonix writes them as cards beside the speech transcript, so one page shows what was said and what was seen.
Which subscription plans include visual transcripts?
Every subscription plan includes them: Core, Advanced, Pro, Enterprise, Medical and Legal. The free trial and pay-as-you-go accounts transcribe speech only.
Which files does it work on?
Video files. An audio-only upload gets its speech transcript as usual, with no visual cards. Files with redactions applied are not analyzed.
Does it identify people or faces?
No. Descriptions cover what is visible, such as a person in a red jacket stepping out of a car. Sonix does not recognize faces, does not name anyone, and does not infer intent, emotion or any legal conclusion. Every description is editable, and the video is one click away to check it.
Where do the descriptions show up?
In the editor, in search, in the plain-text export, in AI analysis and summaries, and through the MCP server. They are not added to subtitles, Word or PDF exports, translations or shared links.
How do I turn it on?
Open your upload settings and enable visual transcripts. A switch then appears on the upload form whenever you add a video. For a video you already transcribed, choose Generate visual transcript in the editor's Tools menu.
Pages around this one
The feature this page belongs to, and its neighbors.
Everything else Sonix does
with the same recording
Get started
Read your next video
Upload a video on any subscription plan and switch on the visual transcript.
4.98
10xfaster than real time
30free minutes to start
