Video analysis for EMCA research
EMCA Vision is a Mac app for ethnomethodological and conversation analytic research on video. It finds the participants in a recording, maps their positions, formations and gaze, and lines them up with your transcript, so you can return to a moment and follow an interaction across talk, gaze and space. Everything runs on your own computer.
The LEGO building task (Davidsen, Thomsen & McIlvenny, 2020): five cameras, three of them 360°, every person found and followed, and the Jeffersonian transcript beside the video.
What it does
Body and face points for every person in one or more cameras, including 360° video. People are matched across cameras and moments, and you can merge, remove or add participants by hand. Everything else builds on these labels.
Import a DOTE transcript, subtitles or a speech recogniser's output. Link speakers to participants (MuVAP, an audio-visual speaker model, suggests the links by watching faces and listening), see turns, pauses, gaps and lapses, and where the speaker looked as the next speaker took over.
The participants seen from above, with their orientation, the groups they form around a shared space (vis-à-vis, L-shape, side-by-side, circle) and how these change over time.
Head direction for every participant, mutual gaze, what or whom they look at, and moments of smiling and laughing, as candidates on a timeline.
Keep several transcripts of a recording, sync them to the video by sound or by ear, and compare two side by side. A draft transcript can be made with Whisper and MuVAP and exported to DOTE to correct.
Save your place, transcript and analysis as a project, export it to a colleague, and reconnect the video on their Mac. Every change can be undone.
Whiteboards and other shared surfaces, walk-ups and pointing at them, and objects people pick up, show and hand over.
Recordings never leave your computer. The analysis models are downloaded once, from their publishers, and checked before use.
Approach
EMCA Vision does not do the analysis. It proposes candidates: moments where gaze shifts, formations change or the floor passes to someone else, so that the analyst can find, compare and return to them faster. Every proposal is marked as exploratory and unverified until you have reviewed it, and the review is kept and can be exported. The rules behind each candidate are written down in plain language, with the thresholds used.
Download
Cite
Davidsen, J., & Paulsen, L. (2026). EMCA Vision (Version 0.8) [Computer software]. Aalborg University.
EMCA VISION@software{davidsen_paulsen_emca_vision,
author = {Davidsen, Jacob and Paulsen, Lucas},
title = {EMCA Vision},
year = {2026},
publisher = {Aalborg University},
url = {https://codeslayer84.github.io/emca-vision/}
}Team