Skip to content
refs
REF-2778Source onlineBlueprint rev 01

ChatGPT Voice at the Table

A staged ensemble launch film turns ordinary conversation into three escalating voice demos, with a playful crew reveal and a late title card.

Maker @OpenAIFor OpenAIOriginal post ↗
Cut timeline · 79 shots03:34
00:0000:3001:0001:3002:0002:3003:0003:30

Field notes / why it works

The 0:00 close-up of knitting hands poses a small human problem, then delays the product announcement until about 0:18. Three demonstrations at 1:03, 1:29 and 2:28 escalate from interruption handling to concurrent fact checking and translation, with reaction shots and jokes making each capability easy to grasp. The 79-shot edit sustains a 3:34 film while a consistent table setting and phone insert keep it legible.

01

Plate 01 / The format in one breath

A polished ensemble comedy begins as an ordinary conversation about knitting. At 0:18 it reveals a voice product, then uses three increasingly demanding, lightly funny conversations to prove turn taking, practical reasoning and live translation. A 0:15 glimpse of the crew exposes the constructed set and makes the ad feel playful.

Why it works

  • The first 2.75 seconds stay on hands and red yarn while a specific domestic question plays; the product name waits until roughly 0:18. The viewer has a scene to follow before the announcement.
  • Each claim is immediately acted out: interruption and handoff at 1:03–1:15, concurrent fact check, transit and weather questions at 1:29–2:22, translation in a mock negotiation at 2:28–3:07.
  • The recurring table, phone and three contrasting conversational archetypes make 79 shots feel coherent; the final 3:24 title reunites them before the 3:30 logo.
02

Plate 02 / Format card

Reference
Platform · aspect · lengthX · 16:9, 1280×720 · 3:34.1 at 23.98 fps
Pace79 shots; 21.9–22.7 cuts/min; 26.9 picture changes/min; mean 2.71 s, median 2.34 s; longest 13.72 s
Script503 transcribed words; 143 wpm overall, 184 wpm in active speech; 72 short sentences averaging 7 words; 15 questions
VoiceSeveral warm, animated older adult conversational voices plus a clear, responsive product voice; timing and interruptions carry the appeal
On screenPredominantly staged live action at one table (inferred >85% of runtime): medium/wide presenters, phone inserts, reaction close-ups; brief slate, titles and logo
CaptionsNo continuous social caption system visible; small conventional white dialogue subtitles appear in a few frames, e.g. 1:09
SoundContinuous tonal/music-like bed detected; dialogue prominent; cut accents do not establish a repeated SFX pattern; −14.9 LUFS, −1.3 dBFS true peak
StructureDomestic cold open → self-aware announcement → three demonstrations → ensemble payoff and brand card; no spoken purchase CTA

Pro / 13 plates + recipe

The first plates are open. The full dissection is Pro.

Get the timed shots, type and sound specs, and the runnable recipe.

Open the full blueprint
  1. Plate 03 / The hook, frame by frame (0:00–0:03)Pro
  2. Plate 04 / Structure (beat sheet)Pro
  3. Plate 05 / Shot grammarPro
  4. Plate 06 / Visual style guidePro
  5. Plate 07 / Voice and deliveryPro
  6. Plate 08 / Sound designPro
  7. Plate 09 / PackagingPro
  8. Plate 10 / Script templatePro
  9. Plate 11 / Production planPro
  10. Plate 12 / PromptsPro
  11. Plate 13 / Edit recipePro
  12. Plate 14 / QA checklistPro
  13. Plate 15 / Make it yoursPro
  14. Recipe / runnable promptPro