Quiz feedback clips with an AI avatar: right and wrong answers

A quiz needs the learner to hear why an answer is right or wrong, and a short clip from the same presenter is one way to do it. Sume does not run quizzes, score answers or branch video; it renders clips. Your LMS or page decides which clip plays. The sensible build is two small clips per question: a correct-answer clip and a wrong-answer clip.

The arithmetic

The shortest clip the route accepts is 4 seconds, and Sume's estimate has a floor at that length. At 10 seconds each, ten questions with two clips each is 20 clips and 200 seconds of video.

Write the wrong-answer clip first

The wrong-answer clip is where most training value sits. Name the specific misunderstanding, restate the correct rule in one sentence, and point back to the lesson timestamp. Avoid wording that blames the learner. The correct-answer clip can be shorter, since it confirms and moves on.

Have a subject-matter expert approve each script. The avatar will say wrong content as fluently as right content. After rendering, check the transcript against the approved text before loading the clip into the quiz.

Keep a naming scheme

Because you may re-render a single question after the policy changes, name your own records by question id and outcome (q07-correct, q07-wrong), and store the job and video ids with them. The tracking-table post has a starting schema.

Limits that shape the design

One final video uses one resolved avatar, so keep a single presenter for the whole quiz. The output is 720p. Sume's Avatar 1.0 speaks English only. A clip cannot adapt to a learner's actual wrong choice beyond the scripts you write, so cover each distractor with its own wrong-answer clip when the distractors reveal different misconceptions; that multiplies the count but not the per-clip price.

Cutting the count

If 20 clips is too many, render a single generic correct-answer clip for the whole quiz and write specific clips only for wrong answers. Ten wrong-answer clips at plus are 24.50 dollars. You can also record clips at the 4-second minimum for simple confirmations, and spend longer on the concepts that learners miss most often.