What NoteGPT Bundles Into One AI Learning Login

Chapter 1

One Login Behind Twenty Different Tools

A graduate student signs up for NoteGPT after a classmate mentions it turns hour-long lecture recordings into notes overnight. She expects a smarter version of a notebook app. What she finds behind the login is closer to a workshop than a notebook: transcription, translation, image generation, voice cloning, video creation, mind maps, flashcards, and an AI chat that can be pointed at half a dozen different language models depending on the question.

The site's own terms describe the underlying idea in one sentence – users upload or input text, files, images, links, audio, video or voice samples, and the service processes them into summaries, transcripts, translations, images, audio, video, slides, notes, mind maps and charts. That is a wide claim, and unlike most wide claims it holds up: every one of those outputs has its own dedicated tool inside the dashboard, not a single generic button trying to do all of them badly.

The marketing language around it stays narrow on purpose. Pricing-page testimonials talk about lecture reviews and video research, and the help page opens by calling itself an AI-powered note-taking tool for clever learning. A student reading only the homepage would reasonably conclude NoteGPT is a summarizer with some extras attached. A student who opens the sidebar finds a content-processing engine wearing a student's backpack, and most of what lives in that engine is never mentioned on the page that sells it.

It also means the tools are under-sold by category rather than by feature. Passport photo cropping and AI-image detection sit in the same dashboard as lecture summarization, filed under separate menu groups a study-focused visitor would have no reason to open. None of that breadth is free in the ordinary sense, though individual clicks are made to feel free. What decides whether a click costs nothing or costs something is a two-currency system running quietly under every button – and understanding it is the first real decision a new account has to make.

What you can actually do here

The list below moves from the tools every visitor eventually finds to the ones filed several menus deep – each row names the chapter that explains it properly, so you can jump straight there.

Turning Raw Material Into Usable Output

Keeping a Character or a Voice the Same Twice

Documents, Verification and Spending Less

Chapter 2

Two Kinds of Quota Running Under Every Feature

Underneath every tool in NoteGPT's dashboard sits one of two meters, and the site names them plainly: Basic Quota and Premium Credit. Basic Quota pays for text-heavy, comparatively cheap work – summarizing, translating, chatting with a base model, generating a mind map. Premium Credit pays for anything leaning on heavier computation – higher-accuracy transcription, speech and podcast generation, image generation, video generation, and chatting with a named frontier model instead of a base one.

The split is not cosmetic. The quota policy page lists a separate cost for nearly every action on the platform, down to the individual chat model: a message to a lighter reasoning model spends a small handful of Premium Credits, the same message sent to the most capable model on the roster spends several times as many. A summarization job priced by word count behaves nothing like an image job priced per output, and neither behaves like a video job priced by the second.

A handful of items also break the basic pattern by design. Saving a cloned voice for later reuse spends a strikingly large slice of Basic Quota compared with the cost of cloning or designing that same voice in the first place – a specialty price sitting inside what looks like the cheap meter.

This matters because a Basic Quota allowance running out and a Premium Credit allowance running out are not the same event. Every paid tier gives an unlimited allowance of its stated basic features, which means transcription, summaries, mind maps and base-model chat effectively stop being rationed once you are on any paid plan. Premium Credit is the one meter that stays finite no matter which tier you choose – it simply starts from a different number, which is why comparing plans by price alone misses the point almost entirely.

Chapter 3

Four Plans That Don't Scale the Way They Look

NoteGPT sells four subscription tiers, and lined up side by side they do not form a neat ladder. The entry-level plan carries a full allowance of Basic Quota but effectively no monthly Premium Credit, which means everything backed by heavier computation – premium transcription, image generation, video generation, chatting with a named frontier model – sits outside what that plan actually pays for, even though the plan's own feature list still names those tools.

One tier holds the site's largest Premium Credit allowance by a wide margin – large enough that the pricing page markets it directly at production-level use and states that four out of five paying customers choose it. Two further tiers sit between the cheapest and the richest, each trading a smaller credit allowance for a lower monthly price, and each capping how many YouTube videos or local files can be processed at once – a ceiling the entry plan and the priciest plan don't share in the same way.

Annual billing is discounted against monthly billing across all four tiers, and an annual subscription auto-renews into a second year by default unless it's cancelled before the renewal date – the same default applies to monthly plans renewing into the next month, with no second confirmation asked for at the point of charge.

For a team rather than a single account, a separate part of the pricing page splits the same three paid tiers per person and drops the per-person rate further as more people join a shared purchase – a way of buying NoteGPT the main pricing toggle never surfaces on its own.

Chapter 4

Locking a Face Across an AI Video With Seven Reference Slots

A creator builds a serialized explainer series and wants the same on-screen host to appear in every episode. The first AI-generated clip looks right. The second clip, generated from a fresh prompt, gives the host a different jawline and a jacket that has quietly turned from leather to denim. This is the character-drift problem that has followed AI video since the first text-to-video models, and it is the specific problem NoteGPT's Gemini Omni agent was built to answer.

Instead of one reference image, the tool provides a matrix of seven reference slots, and the site's own guidance recommends filling them with a genuinely varied set: a clean front-facing shot, a 45-degree three-quarter view, a side profile, and an expression shot showing an emotion. Populating those slots gives the underlying model several angles of the same face to check future frames against, rather than one flat photograph.

This sits alongside a related tool: an AI Character Description Generator that turns a single line of an idea into a structured profile covering appearance, personality, background and abilities. That written profile then feeds directly into an AI Character Generator, which renders it in one of nine named visual styles, from realistic through anime to a specific dark-gothic option.

Together the two tools solve slightly different halves of the same complaint. The reference-slot system keeps a face consistent across generated video frames; the description generator keeps a written character consistent across separate image generations, so a character regenerated next week still matches today's version. Neither replaces careful prompting, and both depend on the same quiet discipline – writing the character down once, before asking the model to draw or animate it.

Chapter 5

Voices, Faces and Documents Built to Pass Inspection

Somewhere between the summarizer and the video tools sits a smaller cluster of features built for a narrower kind of task – making a voice, a face or a document pass a specific, external check rather than simply look good on screen.

The voice side covers text-to-speech, podcast generation, voice cloning and prompt-based voice design, plus a real-time multi-voice mode that assigns different voices to different speaking roles inside one script – useful for a scripted interview or a two-character dialogue rendered without separate exports. The voice library spans more than a hundred voices across multiple languages, by the site's own count.

The document side is smaller but pointed: an AI Passport Photo Maker that detects a face, crops to a chosen document size, and swaps the background to a plain colour, plus an optional feature that can digitally change the clothing someone is wearing in the photo. That last option carries a real catch, covered in the next chapter's limits.

A third tool in this cluster runs the opposite direction – an AI Image Detector that takes an uploaded image and returns a percentage likelihood that it was AI-generated, along with a written breakdown pointing at specific evidence: fur texture, light falloff, shadow irregularity, the kind of detail nobody consciously checks when glancing at a photo. It exists for the opposite worry from everything else in this chapter – not whether something will pass as real, but whether you just got shown something that isn't.

Chapter 6

Where NoteGPT's Free Tier Stops and the Refund Clock Starts

NoteGPT is free to use in a genuine sense – the help page states it outright – but free here means a metered allowance rather than an unlocked product. The free plan hands out a small amount of Basic Quota and effectively no Premium Credit, enough to try most tools once or twice and not enough to build a routine around any of them.

Refunds run on a short, strict clock. NoteGPT will refund a purchase only within 24 hours of the very first time an account pays for anything – not a renewal, not an upgrade, not a repurchase after cancelling once before. Even inside that window, the refunded amount is calculated as the payment minus the value of any Premium Credits already spent, so testing every premium tool immediately after upgrading and asking for a refund the next morning returns less money than the plan cost.

Cancellation and refund are two different actions with two different outcomes. Cancelling a subscription through the account settings stops the next renewal charge but doesn't return any money already paid – the account keeps its current tier's access until the paid period actually ends, then drops to the free plan. A refund, by contrast, ends access immediately and only applies inside that first-purchase window described above.

A few categories are excluded from refunds entirely regardless of timing: any purchase made at a discounted or promotional rate, and any lifetime Premium Credit purchase. Read together, the refund and cancellation rules reward one thing specifically – deciding before the first purchase rather than after it, because the safety net only exists for the very first transaction and only for one day.

Chapter 7

What the Terms Decide About Who Owns the Output

Every AI platform eventually has to answer an odd legal question: who owns something a model generated from someone else's prompt? NoteGPT's terms answer it in the reader's favour, then immediately add a qualifier most users never read as far as.

The terms grant full ownership of generated output to the person who requested it, with rights to use, modify, distribute and commercialise it, and no requirement to credit NoteGPT unless separately agreed. The qualifier sits one clause later: because the underlying models are shared across every account, an output may not be unique, and other users may receive a similar result from a similar prompt. Ownership of your output does not extend to whatever another user's account produced, even if it looks close to identical.

A second boundary worth knowing sits in the liability section rather than the ownership section: NoteGPT's total responsibility for any claim is capped at whatever that account paid in the twelve months before the claim arose, and the service carries no promise of uninterrupted or error-free operation. That's a standard SaaS position, not a distinctive one, but it's the ceiling under everything else described in this book.

The company also spends an entire dedicated page warning that it is the only legitimate NoteGPT – naming its Chrome and Edge extension listings specifically, and stating flatly that it will not support accounts created through any unofficial clone. For a platform built around uploading transcripts, drafts and voice samples, the distinction between the official domain and a look-alike is not a footnote; it is the first security decision a new user actually makes, usually without realising a decision was being made at all.

Questions readers actually ask