Custom Interview Avatar: Creation Guide
How to record, submit and activate a custom avatar for your JobMojito interviews.
Written By Jozef
Last updated 27 days ago
1. Overview
A custom avatar lets you run JobMojito interviews with a face from your own organisation instead of one from our standard avatar library. The avatar speaks, listens and responds in real time during candidate interviews, driven by the JobMojito interview engine.
The process has three phases:
You do not need any technical setup. You send us the recording, we handle the build, and the avatar appears in your JobMojito account ready to assign to interview templates.
2. Before you start
Confirm with your JobMojito account contact:
That custom avatars are included in your plan and that you have an available avatar slot.
The delivery format and resolution agreed for your account.
Who will appear as the avatar, and that this person is available for the recording session.
Choose the right person. The avatar becomes the face of your interview process for every candidate who applies. Common choices are a talent acquisition lead, a hiring manager, or a brand-appropriate spokesperson. Consider how long that person is likely to stay with the organisation, since replacing an avatar means repeating this process.
3. Recording the footage
3.1 Structure
This is the part that most affects the final quality. The recording must be one continuous take, with no cuts or edits, and at least 2 minutes long. It contains three segments recorded back to back.
The distinction between segments 1 and 3 matters. Listening is animated and responsive. Idling is calm and static. If the two look the same, the avatar will behave oddly between interview questions.
3.2 Script for the talking segment
The words themselves are not reused in interviews, so the content is not critical. What matters is natural, well paced delivery with varied intonation. A short introduction to your company and hiring process works well and keeps delivery authentic. Avoid reading in a flat monotone or over-rehearsing to the point of stiffness.
3.3 Background and environment
Clean, static background, or a professional grade green screen.
No moving people or objects, no shifting light, no reflective surfaces behind the subject.
If using a green screen, raise it off the floor to reduce colour spill.
Keep the subject at least 1.5 m (5 feet) clear of the backdrop to avoid cast shadows.
3.4 Camera and framing
Minimum 1920x1080 (HD). Higher resolutions such as 4K are not used, because real time interview streaming requires a lower latency format.
Frame the subject chest up. Full body recordings are not supported.
Keep hands out of frame where possible. If visible, they should stay still and relaxed.
Fix the camera on a tripod. No handheld footage, no zooming, no repositioning mid-take.
3.5 Lighting
Soft, even lighting with no harsh shadows on the face.
Two diffused lights placed either side of the subject.
Optional backlight above and behind the head to separate the subject from the background.
Slightly warm colour temperature (around 4800K) to keep skin tones natural.
If shadows appear under the chin, add a soft fill from below.
3.6 Direction for the person on camera
Maintain steady eye contact with the lens for the entire take.
No swaying, no audible breathing, no fidgeting during the silent segments.
Remove glasses, jewellery and shiny or patterned clothing. These interfere with background separation and can produce visual artefacts.
Plain, solid colour clothing works best. Avoid green if using a green screen.
Record two takes: one with minimal movement, and one with 4 to 5 slow, subtle gestures kept below shoulder height. We will select the stronger source.
4. Consent
Every custom avatar of a real person requires a recorded consent statement from that person. This is a firm requirement and submissions without it cannot be processed.
Requirements:
The consent recording must feature the same person as the training footage.
It must be recorded and submitted at the same time as the footage, not retroactively.
The person must state their name and confirm that they agree to a digital likeness being created and used.
If you are producing the avatar on someone else's behalf, that person records their own consent. They do not need to be in the room, and we can provide a link or QR code so they can submit directly from their phone.
What we cannot accept:
Footage of a person who has not personally given consent.
Stock footage, or footage of public figures.
AI generated or synthetic source footage presented as a real person.
Consent records are retained as part of your account's compliance file and are available on request for audit purposes. See the JobMojito compliance documentation for how avatar consent is handled under GDPR and the EU AI Act.
5. Submitting your recording
Recordings are too large for email, so upload them to any file storage service you already use (Google Drive, Dropbox, OneDrive, WeTransfer or similar) and send us a shareable link.
Send the link to support@jobmojito.com with the following:
Both recording takes, unedited, in their original resolution.
The consent recording.
The name you want the avatar to be listed under in your account.
Any voice preference (see section 7).
Before you send:
Check the sharing permissions on the link. Restricted links that require an access request will delay processing. Set the link so that anyone with it can view, or grant access to support@jobmojito.com directly.
Confirm the link does not expire before we have processed the files.
Name the files clearly, for example
companyname-take1.mp4,companyname-take2.mp4,companyname-consent.mp4.Do not compress, trim, colour grade or otherwise process the footage. We need the original files.
We confirm receipt and validation results by email. If you have not heard from us within one business day, check the link permissions first, then follow up.
6. Build and validation
Once we receive a complete submission:
We validate the footage against the requirements above and flag any issues before processing starts.
The avatar is built and quality checked.
The avatar appears in your JobMojito account, ready to assign.
Typical turnaround is 1 to 2 business days from a valid submission. Submissions that fail validation are returned with specific notes on what to refilm.
Rebuild policy: you can request one rebuild from the same footage at no cost if the result does not meet expectations. Rebuilds from new footage count against your avatar slot allowance. Contact support@jobmojito.com to request one.
7. Voice
The avatar's voice is configured separately from its appearance. You have three options:
Matched voice. A voice profile generated from your recording, so the avatar sounds like the person on camera. Requires clean audio in the talking segment.
Library voice. Selected from our voice library, chosen per language.
Neutral default. Our standard interview voice.
For multilingual hiring, voice is assigned per language rather than one voice covering all of them. If you interview across several markets, discuss voice mapping with your account contact during setup.
8. Activating the avatar in interviews
Once the avatar is live in your account:
Open the interview template you want to use it with.
Select the custom avatar from the avatar picker.
Run a preview interview to check appearance, voice and pacing before publishing.
Publish the template.
The avatar's questions, scoring logic and conversational behaviour come from your JobMojito interview configuration, not from the recording. Changing what the avatar says never requires new footage.
9. Troubleshooting
10. Pre-shoot checklist
Before the session
[ ] Avatar slot confirmed with your account contact
[ ] Person on camera identified and briefed
[ ] Script prepared for the 90 second talking segment
[ ] Camera on tripod, set to 1080p, framed chest up
[ ] Background clean and static, subject 1.5 m clear of it
[ ] Two key lights, optional backlight, positioned and diffused
[ ] Glasses, jewellery and patterned clothing removed
[ ] Room quiet, no background noise
During the session
[ ] Take 1 recorded: minimal movement, all three segments, no cuts
[ ] Take 2 recorded: 4 to 5 subtle gestures, all three segments, no cuts
[ ] Consent recording captured in the same session
[ ] Footage reviewed on a large screen before wrapping
After the session
[ ] Original unedited files uploaded to your file storage
[ ] Share link permissions checked
[ ] Link, avatar name and voice preference sent to support@jobmojito.com
[ ] Receipt confirmation received from JobMojito
[ ] Preview interview run before publishing to candidates
Questions
For anything not covered here, contact support@jobmojito.com.