Glossary
Melda has a small set of repeated terms that show up across the product: nuggets, swipe files, hooks, vibes, voice profiles. This page is the plain-language reference.
Each term gets a single, quotable definition — what it is, what it does, and why it matters in a creator workflow.
Every term traces to a real surface in Melda. The glossary explains the language the rest of the product uses.
Definitions are sized to be answer-engine quotable, with no jargon hiding behind brand-specific shorthand.
Definitions
Read top to bottom for the natural creator flow — first you save examples (swipe file), then you extract a nugget, then you choose a hook and a vibe, and the whole thing is conditioned on a voice profile.
A bounded teachable idea extracted from a creator transcript or source material. Each nugget is one repurposable script-fuel unit: a clear claim, supporting evidence from the source, and the specific move the creator is making. Melda extracts nuggets so generated scripts have something real to say instead of starting from a blank prompt.
A personal library of content examples a creator wants to learn from — strong hooks, vibes, structures, moves seen in the wild. The point is a durable taste library you can pattern-match against later instead of starting every script from a blank prompt.
The opening line of a short-form script. The hook bar is absolute: would a human actually find it interesting and want the next sentence? Strong hooks come from tension, contradiction, status threat, consequence, unusual specificity, or a believable identity-state shift the viewer wants to inhabit.
The tonal and structural register of a short-form script — punchy, mentor, calm diagnostic, warm authority — that determines how the underlying nugget gets delivered. Melda picks a vibe that matches both the nugget evidence and the creator voice profile so the script feels native rather than templated.
A captured signature of how a creator actually talks and writes — tone, vocabulary, teaching style, signature phrases, argument structure — extracted from their own content. Melda uses voice profile as conditioning context for every generated script so output reads as spoken proof in the creator voice.
A word-for-word spoken script written for a short vertical video — typically 45 to 90 seconds of a creator talking directly to camera on TikTok, Instagram Reels, or YouTube Shorts. Unlike an outline or bullet points, a short-form script is meant to be read aloud as written.
Turning one piece of source material — a podcast episode, long video, coaching call, or interview — into several new publishable assets. For creators, the highest-leverage form is finding the strongest bounded ideas inside existing recordings and reshaping each one for a specific format rather than reposting the same clip everywhere.
A video where the creator speaks directly to the camera with no b-roll, edits, or effects carrying the content. It is the lowest-production, highest-trust short-form format, and the format every Melda script is structured for: if you can read it aloud, you can post it.
One of several candidate opening lines generated from the same underlying idea. Because the hook decides whether anyone hears the rest of the script, generating variations lets the creator choose the tension, framing, or specificity level that fits their audience before recording.
A tool that produces spoken-word scripts from source input. Source-grounded script generators like Melda differ from general-purpose AI writers: they start from the creator’s own recorded material and extracted ideas, so every script has real proof behind its claims instead of plausible-sounding filler.
The constraint that every generated script traces back to something actually said or shown in the creator’s own material. Source grounding prevents hallucinated quotes and invented facts: if the source does not support a claim, the honest result is fewer ideas, not fabricated ones.
