How to turn a raw consultation recording into a case study video

Last updated 2026-08-18 · 6 min read

What is the short version?

A recorded consultation is the most valuable footage a practice has and the least usable. The room is noisy, the framing is wrong for a feed, and the story is buried in forty minutes of talking that nobody outside the room will sit through.

1. Go to Vizard Agent and upload the raw recording.
2. Say what the case study demonstrates and who it is for.
3. Say the length, the shape and how much cleanup the audio needs.

What do you need before you start?

The recording and the story you want it to tell. Vizard Agent measures the audio and tests the framing itself, so what only you can supply is the argument the footage is being cut to make, since a consultation contains several possible stories and only one of them is the case study.

• The raw recording. Untrimmed. Vizard Agent works better with everything.
• The story. What this case demonstrates, in one sentence.
• The consent position. Whether the person in it has agreed to this being published.
• The length. Four minutes is long for a feed and right for a case study.
• The shape. Vertical for social, and the source is usually wide.

What do you type into Vizard Agent?

Describe the finished piece, not the edit. Vizard Agent reads a brief written for a human editor as instructions, so a paragraph naming the story, the cleanup and the format gets you further than a shot list, and Vizard Agent will handle the sequencing from the transcript.

Prompt

Variants worth knowing:

• A short cut too. Ask for a one-minute version alongside; it comes from the same analysis.
• Names removed. Say so if the person should not be identifiable by name.
• Captions throughout. Almost always worth it for a professional audience watching muted.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real case study. Two stretches stand out: it looked at a picture of the noise before deciding how to clean it, and it rendered both crop options rather than choosing one from a thumbnail.

1. Probes the recording and reviews the project's history.
2. Extracts sample frames and transcribes the consultation audio.
3. Measures the framing with a coordinate grid and inspects the overlay to plan the crop.
4. Reads the transcript in full, in three passes, and checks the word-level timings.
5. Downloads the footage and measures the audio levels, the overall loudness and the silent regions.
6. Analyses the noise floor and the spectrum, then looks at the spectrum image.
7. Verifies the key lines and checks the practitioner's name rather than repeating it blind.
8. Tests crop windows across the whole video and reviews the candidates.
9. Renders both framings at full size, compares the standard and the close-up side by side, then tests the noise cleanup chain and measures the speaker's level before and after.

Step 6 is the one nobody expects. Choosing a noise reduction by ear on a laptop is guesswork; looking at where the noise actually sits in the spectrum tells Vizard Agent what to remove and, more importantly, what to leave alone.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 243.9 seconds, AAC audio. Vertical, just over four minutes, reframed from a wide recording, with the room noise reduced and the speech levelled.

Four minutes is unusually long for a vertical video and correct for this format. A case study is watched by people evaluating a service, and they will sit through length that a casual scroller would not, so Vizard Agent paces it for attention rather than for the first three seconds.

When does this not work well?

A consultation is a private conversation before it is footage. Vizard Agent cleans it, cuts it and frames it, and every question about whether it should be published at all sits outside what the tool can judge, which in this format is the part that actually decides whether the video ships.

• Consent is not implied by recording. Agreeing to be recorded and agreeing to appear in marketing are separate permissions.
• A case study is a claim. What the video says about outcomes is regulated in most professional fields.
• Noise reduction has a cost. Pushed hard, it thins the voice. Vizard Agent measures before and after so the trade is visible.
• A wide room crops badly to vertical. If two people are in shot, one of them is leaving the frame.
• The story has to be in the recording. Vizard Agent can find it and cannot supply what was never said.

How do you fix a result that came back wrong?

Say which part. Vizard Agent keeps the transcript, the word timings, the noise analysis and both rendered framings, so changing the crop or easing the cleanup is a re-render against measurements it already made rather than a fresh pass over four minutes.

• "Use the wider framing." Already rendered; a straight swap.
• "The voice sounds thin." Less reduction, measured against the same noise profile.
• "Cut the section about pricing." Removed and re-paced from the transcript.

How does Vizard Agent compare to doing it yourself?

By hand this is watching forty minutes twice, marking the usable lines, guessing at a noise reduction, cropping to vertical and discovering at minute three that the subject drifts out of frame. Vizard Agent front-loads the measuring so those discoveries happen before the cut, not after it.

Common questions

How long should a case study be? Three to five minutes. The audience is evaluating you, so they will watch longer than a casual viewer.

Can it remove the client's name? Yes. Say so and Vizard Agent cuts around it and keeps it out of the captions.

Will the audio actually be usable? Vizard Agent measures the noise and the speech level rather than applying a preset, which is what keeps the voice intact.

Can I get a short version for a feed? Yes. Ask for both and the short cut comes from the same transcript and analysis.

What if two people are on camera? A vertical crop fits one. Say which, or take the wide version instead.

Does it need the whole recording? Yes, ideally. Vizard Agent finds the story by reading all of it rather than a section you pre-selected.

Can it add captions? Yes, and for this audience it is usually worth it. Vizard Agent cuts them from the same word-level transcript it used to build the edit, so they land on the syllable rather than drifting.

Can I reuse the analysis for a second case study? Within the same project, yes. Vizard Agent keeps the transcript and the noise profile, so a second cut from the same session starts from the measurements rather than from the raw file again.