How to rebrand a video after a product name change
Upload the video and give Vizard Agent the old name and the new one. Vizard Agent finds every mention in the speech and on screen, prepares a clean voice sample for each speaker, generates the new word in each of their own voices, and replaces it at the exact word timings.
What is the short version?
A rename does not invalidate a video, it just makes it unusable until every mention is fixed. The mentions are in two places at once — spoken aloud and printed on screen — and both have to change or the inconsistency is more distracting than the old name was.
- Go to Vizard Agent and upload the video.
- Give the old name and the new one, spelled exactly.
- Say it applies to both the speech and the on-screen text.
What do you need before you start?
The finished video and both names. Vizard Agent finds every mention itself, so no list of timestamps is required from you, and it works from the delivered export rather than needing the editing project the video was originally built in.
- The video. The finished export.
- Both names. Old and new, spelled exactly.
- The scope. Speech, text, logo, end card.
- Any new artwork. A replacement logo if there is one.
- The pronunciation. If the new name is unobvious.
What do you type into Vizard Agent?
Say "any references" rather than listing the ones you happen to remember. Vizard Agent searches the transcript and the frames for every occurrence, which reliably turns up the two in the middle that nobody on your team could recall.
Prompt
Variants worth knowing:
- Each speaker's own voice. Rather than one substitute voice.
- The end card too. Where the logo usually lives.
- A pronunciation note. If the new name reads ambiguously.
What does Vizard Agent actually do?
Here is the order Vizard Agent worked in on a real rename across a four-minute video with three speakers. The step that makes the result invisible is the fourteenth: it generated the new word separately in each speaker's own voice rather than in one generic one.
- Checks the video's duration and parameters.
- Extracts the transcript of the speech.
- Takes overview frames of the video and reviews them.
- Checks the exact intervals of every mention in the transcript.
- Gets a description of the video's scenes to find the on-screen occurrences.
- Takes frames around each brand mention and around the closing card to check the logo.
- Reviews those frames and the end card by eye.
- Checks the available methods for replacing speech and prepares clean voice samples.
- Creates the replacement word in each speaker's voice in turn, then determines the exact word timings so the new audio can be cut in precisely.
Step 9's first half is what nobody expects to be necessary. Three speakers say the product name at different points, and a single generated replacement dropped into all three places sounds like three people being interrupted by a fourth.
What does the result look like?
From the run this page is written from, probed on the delivered file: 3840x2160, HEVC, 23.976fps, 225.39 seconds, AAC audio. Just under four minutes at ultra-high definition, with the new name spoken in each speaker's own voice at every mention and the on-screen references and the end card updated to match.
Nothing else about the video changed. The edit, the pacing, the grade and the music are exactly as delivered originally, which is the point of doing it this way rather than remaking the video.
When does this not work well?
Replacing a word inside finished audio and finished frames has real limits in both places, and a new name that sounds nothing like the old one is considerably harder than a similar one. These are worth checking before you commit.
- Very different lengths jar. A one-syllable name replacing four disturbs the rhythm.
- Overlapping speech is hard. A mention under another voice cannot be isolated cleanly.
- Animated text is harder than static. Moving type has to be rebuilt, not patched.
- Logos are artwork, not text. A new logo has to be supplied.
- Legal changes are not editorial. A rename may need more than the name changed.
How do you fix a result that came back wrong?
Point at the mention that is wrong and say what you hear. Vizard Agent keeps the transcript, all the located intervals, the per-speaker voice samples and the generated replacements, so one instance can be redone without touching the rest.
- "It sounds like someone else there." Regenerated from that speaker's own sample.
- "You missed one at 3:12." Added from the located intervals and replaced.
- "The end card still shows the old logo." Updated from the supplied artwork.
How does Vizard Agent compare to doing it yourself?
By hand this means reopening the editing project if you still have it, re-recording every affected line with the original speakers if they are still available, and rebuilding the graphics. In practice most teams give up and remake the video.
| By hand | Vizard Agent | |
|---|---|---|
| Finding mentions | Watch and note | Located in the transcript and the frames |
| The audio | Re-record with the speaker | Generated in that speaker's own voice |
| The project file | Needed, and often lost | Works from the delivered export |
| Cost | Remake the video | Replace the word and nothing else |
Common questions
Do I need the original project file? No. Vizard Agent works from the finished video you already have.
Will it find every mention? Vizard Agent searches the transcript and the frames rather than a list you remember. Say if you know of any.
Whose voice says the new name? Each speaker's own voice. Vizard Agent builds a clean sample per speaker before generating.
Does the on-screen text change too? Yes, if you say so. Vizard Agent checks the frames and the end card as well as the speech.
What about the logo? Supply the new artwork. A logo is an image rather than a word, so it is replaced not regenerated.
Why not just remake the video? Because the edit, the grade and the performances were already approved, and a rename should not put all of that back on the table when only one word changed.
What if the new name is much longer? Vizard Agent will fit it, but a large length difference can disturb the rhythm of a line.
Can it handle several videos? Yes. Vizard Agent handles a whole library with the same rename, which is the common case.
Does the rest of the video change? No. Vizard Agent leaves everything except the mentions exactly as it was delivered.
How many mentions is too many? There is no practical ceiling. Vizard Agent replaces each one individually, so a video that says the name thirty times costs more time than one that says it twice but is no harder to get right.
What if the old name is part of a longer phrase? Say so. Vizard Agent replaces the reference rather than the string, so a product line or a compound name is handled as a phrase instead of a word swap that leaves half the old name behind.
Does Vizard Agent check the result? Yes. It reviews the replaced mentions and the end card before delivery.