AuroraNote.app VS Sing App React Java

AuroraNote.app's answer

Most AI study tools do one step. AuroraNote does the whole chain: a raw lecture recording goes in, and a complete, tested study kit comes out — speaker-labelled transcript, structured notes, executive summary, mind map, flashcards, quiz, an AI tutor over your own documents, and a narrated podcast episode.

Three things are genuinely unusual. Your video never leaves your device: for video files the audio track is extracted inside your browser with a WebAssembly build of ffmpeg and normalised to a small mono MP3, so only audio is ever uploaded. Transcription includes speaker diarization, so a seminar or a Q&A reads like a script instead of a wall of text. And the flashcards are scheduled with FSRS-6 — the algorithm Anki moved to — rather than the SM-2 heuristic from 1987 that most flashcard apps still run.

AuroraNote.app's answer

Because the alternatives each solve one part of the problem and leave you to join the rest by hand.

A general assistant like ChatGPT or NotebookLM will summarise a document, but it will not transcribe a two-hour lecture with speaker labels, schedule your revision, or test you. A transcription tool like Otter gives you accurate text and stops there — text is where studying starts, not where it ends. A flashcard app like Anki or Quizlet has the scheduling right but expects you to write every card yourself, which is the reason most people quit. Other AI study apps cover more ground but upload your lecture video to their servers and schedule cards with SM-2.

AuroraNote is the chain, end to end, and it is stricter than its category on the two points that matter: the video stays on your device, and the spaced repetition uses a modern scheduler.

AuroraNote.app's answer

Frontend: Next.js 16 with React 19 and TypeScript, Tailwind CSS 4, Framer Motion, KaTeX for formulas and Mermaid for diagrams. Audio is extracted client-side with ffmpeg compiled to WebAssembly (@ffmpeg/ffmpeg), and spaced repetition runs on ts-fsrs (FSRS-6). Deployed on Vercel.

Backend: Python with FastAPI, deployed on Railway. Supabase provides Postgres, authentication and storage, with row-level security and pgvector for retrieval. Payments run on Stripe.

AI: Deepgram Nova-3 for transcription with diarization, Mistral OCR for scanned documents and photos of handwritten notes, mistral-embed for retrieval embeddings, DeepSeek for generation, and ElevenLabs Flash v2.5 for podcast narration.

AuroraNote.app's answer

University students who have more course material than time, and whose material is not already in a convenient form.

In practice that means four groups. Students whose lectures are recorded and who never rewatch them. Medicine, law and engineering students facing hundreds of pages of dense reading per exam. Students taking a degree in a second language, who need to attend in one language and revise in another. And anyone who still takes notes by hand and wants them searchable.

It is built for the exam cycle specifically, not for general note-taking: the features that matter most — quizzes with a score, spaced repetition, a tutor that only knows your syllabus — are the ones you reach for in the last two weeks before a paper.

AuroraNote.app's answer

It started from a pile of unwatched lecture recordings. Recording is easy and rewatching never happens: two hours of video cannot be skimmed the night before an exam, and scrubbing back to find the one slide you needed burns the time you do not have. The recordings were the most complete notes available and the least usable thing on the laptop.

The first version just produced transcripts. That turned out to be the wrong finish line — a verbatim transcript is raw material, not learning — so the product grew into the steps that come after: structure it, map it, then test yourself on it. Keeping the video on the device was a decision from the start, because a lecture recording contains a room full of people who did not consent to being uploaded anywhere.

AuroraNote.app's answer

None to name yet. AuroraNote launched in July 2026 and is used by individual students rather than institutions, so there are no organisational customers to list.