← All works
AI Video1 min read 4 sections

AI Virtual Host: A Presenter Who Was Never Filmed

A realistic AI-generated presenter in a podcast studio, built for vertical explainers: natural expressions, steady identity, and lip-sync to the voice.

Format
Vertical video, 9:16
Length
43 seconds
Made with
Generative AI video and voice
Use
Localised educational explainers
  • 43s

    Of continuous presenting

  • 9:16

    Made for phones

  • 1

    Host, consistent shot to shot

  • 0

    Cameras or studio bookings

AI Virtual Host: A Presenter Who Was Never Filmed
0104

Overview

A presenter for educational and technical explainers, generated rather than filmed: a woman in a red saree at a podcast microphone, in a warm studio set. The goal was a host who could present in any language without booking a studio.

AI Virtual Host
0204

Range of Expression

A host who holds one expression reads as a mannequin. These frames, from across the clip, show the range: speaking, smiling, a thoughtful pause, and a relaxed wide shot.

Four moments from the clip
Four moments from the clip
MomentWhat it shows
Mid-sentenceMouth shapes that follow the voice
SmilingWarmth, so it feels like a person, not a reader
A pauseClosed eyes and a breath between thoughts
Wide shotThe same host and set from a second angle
What makes a host believable
0304

What Makes It Hard

ChallengeWhy it matters
Lip-syncViewers notice a mismatch between mouth and voice instantly
Natural blinks and gazeA fixed stare is the fastest way into the uncanny valley
Identity across anglesThe close-up and the wide shot must be the same person
Jewellery, fabric, handsSmall details are where generated people usually break
The craft problems a virtual host has to solve
0404

Format

DetailValue
Aspect ratio9:16 vertical
Resolution720 × 1280
Length43 seconds
Intended useExplainers localised into regional languages
Delivery specs
The opening frame
The opening frame