How to make a short that teaches the difference between two similar-sounding words
Give Vizard Agent the two words and the mistake learners make. It auditions a voice for the explanation and a separate one for the demonstration, tests how the explaining voice handles the foreign word inside its own sentences, and puts the phonetic spelling on screen next to each one.
What do you type into Vizard Agent?
Say the two-voice structure explicitly. A single narrator reading a sentence that switches language for two words is the default, and it is the one thing that ruins this format: the demonstration comes out with the explaining language's accent. Give the mistake too, because a learner saying the wrong word in a real sentence is the hook.
Prompt
What does Vizard Agent actually do?
A pronunciation lesson has one requirement no other video has: the word being taught must be said correctly. A voice built for one language reads a foreign word with that language's sounds, so Vizard Agent hears the mixed reading before building anything else, and re-checks the explanation after any re-record. In the session behind this article it:
- Auditioned a narrator voice in the explaining language, tested how it pronounced the foreign words inside its sentences, and listened to the mixed reading.
- Tested the typeface for the writing system, then generated the explanation and the demonstration narration separately.
- Built motion cards with the words and their phonetic spelling, rebuilt the tongue-twister card after reviewing it, and pulled word-level timings so the cards landed on the words.
- Switched to a clearer static caption style after checking legibility, and checked the loudness difference between the music and the narration.
- Re-recorded the whole narration in the user's own voice from a sample, checked each explanation sentence was still correct, and re-timed the cards.
- Delivered a version with no voiceover alongside the finished one.
When does this not work well?
The format depends on the difference being audible in a short clip played on a phone. Very subtle contrasts don't survive phone speakers, and differences you have to see, like tongue position, need a diagram or a face. The mistake differs by the learner's first language, and accents vary, so say which one you're teaching. One pair per video works; five doesn't.
Common questions
Why two voices? Because one voice reading both languages mispronounces the word being taught. Vizard Agent uses a separate native voice for the demonstration.
Can it use my voice? Yes. Send a sample and Vizard Agent re-records the explanation in it.
Will it get the phonetics right? Vizard Agent looks them up and shows them. Check them; you are the teacher.