Vizard Agent

How to make a vertical sports news recap with a commentator voice

Last updated 2026-08-17 · 6 min read

Tell Vizard Agent the story and the register you want. Vizard Agent searches for a commentator voice that actually fits, records the narration to the length you asked for, sources and generates the visuals, builds animated captions and news badges, and mixes stadium sound under the read.

What is the short version?

A sports recap is thirty seconds of a story people already half know, told with enough energy to be worth watching anyway. The whole thing lives on the voice; everything visual is there to keep the eye busy while the voice does the work.

  1. Go to Vizard Agent and say what happened, in one or two sentences.
  2. Say the voice: accent, register, energy.
  3. Say the length, the shape and the pacing.

What do you need before you start?

The story and the voice. Vizard Agent writes the script, sources the visuals and builds the graphics, so what it needs from you is the factual core — the names, the result, the context — and the register the recap should be delivered in.

What do you type into Vizard Agent?

Give the facts and the style in one message. Vizard Agent writes the commentary from what you supply and sources the visuals itself, so the accuracy of the finished recap comes entirely from the accuracy of the sentence or two you write here — it has nothing else to check against.

Prompt

Create a [30] second vertical [football] news video about [what happened]. Dramatic sports-news style, [British male] voiceover, fast pacing, captions on.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real sports recap. The casting is the most persistent stretch of the run — four separate searches — because a recap with the wrong voice fails no matter what the pictures do.

  1. Checks the generation, voice, music, stock and caption tools available.
  2. Searches the stock library for portrait football footage.
  3. Searches four different ways for the right commentator voice — dramatic, energetic, British, news-anchor.
  4. Generates the commentary, checks its duration, and regenerates twice to land on the target length.
  5. Transcribes its own narration for word-level timings.
  6. Downloads music candidates and generates stadium and impact sound effects in parallel.
  7. Downloads the stock clips and generates custom imagery, then builds a contact sheet of everything and looks at it.
  8. Generates animated captions in two styles, checks the bold fonts and the caption timings.
  9. Mixes the multi-track audio, renders the clips with camera motion, adds transparent news badges — then checks the quality-control frames and re-renders the captions at a better position.

Steps 3 and 4 are the two nobody plans for. Finding a voice took four searches with different wording, and hitting a thirty-second target took three generations — because a script that reads as thirty seconds on the page rarely does out loud.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 29.0 seconds, AAC audio. Vertical, twenty-nine seconds, commentary over sourced and generated visuals with camera motion, animated captions, news badges and stadium sound under the read.

Vizard Agent extracted quality-control frames twice — once before the caption fix and once after — and then ran an automated check over the delivered file.

It is not instant, and this category was not separately measured. Comparable work runs a median of 28 to 38 minutes end to end. Across all projects the median cost by tier is Flash 47, Pro 55, Max 242, Ultra 263 credits.

When does this not work well?

Sports recaps are made of two things Vizard Agent cannot supply: real match footage and verified facts. It writes and produces a convincing recap of whatever you tell it, and both of those limits matter more here than anywhere else.

How do you fix a result that came back wrong?

Say what needs changing and where. Vizard Agent keeps the narration, its measured word timings, every sourced and generated visual and the badge overlays as separate assets, so a corrected fact re-reads a single line and a caption fix re-renders the text alone without touching the pictures or the mix.

How does Vizard Agent compare to doing it yourself?

By hand this is a script, a voice you either record or licence, a stock search, badges built in a design tool, a caption pass and a mix — every day, for a format whose value drops by the hour.

By hand Vizard Agent
The voice One option, take it Four searches until it fits
Hitting the length Read it and hope Generated, measured, regenerated
Captions Type and time them From the narration's word timings
Checking Watch it once QC frames pulled, captions re-rendered

Common questions

Can it use real match footage? No. Broadcast rights make that impossible; Vizard Agent builds around stock and generated visuals with commentary over them.

Does it check the facts? No. Vizard Agent writes the recap from what you tell it, so the accuracy is yours to confirm.

Can I get a particular accent? Yes. Name it and Vizard Agent searches until it finds a voice that matches.

How long should a recap be? Thirty seconds. Anything longer stops being a recap and starts being analysis, which is a different format with a different pace.

Can it caption the recap? Yes. Vizard Agent builds animated captions from the narration's own word timings and checks their position on the frames.

Can I do one every day? Yes. Keep them in one project so the voice, badges and captions stay identical across the series.

Can it use my own clips? Yes, and it prefers them. Upload anything you have the rights to and Vizard Agent cuts around your material first.

Does it add sound effects? Yes. Vizard Agent generates stadium ambience and impacts and mixes them under the commentary rather than leaving the read dry.