Loading…
Loading…
Record a thermodynamics lecture, photograph a whiteboard, or upload a textbook chapter. Distill writes LaTeX-perfect equations, builds concept graphs, and generates fresh problem sets from your own course content.
No credit card required · 10 recordings free every month · 50% student discount with .edu email
Signals & Systems — Lec 12.mp4
1h 15m · Captured for engineers
Engineering lectures are 30% talking and 70% whiteboard. The professor writes Maxwell's equations, says 'clearly,' skips three steps, and expects you to fill them in by Tuesday's problem set. Meanwhile your phone is recording blurry audio that can't see the board.
Distill solves the capture gap for STEM. Upload the lecture audio alongside a photo of the whiteboard and we'll align them — the equation transcription will land at the exact timestamp the professor started writing it. LaTeX comes out clean. Derivation steps are numbered. The three skipped algebraic moves are filled in by our reasoning pass so you can verify them against what the professor actually wrote.
We built the equation layer with former PhD students from electrical and mechanical engineering. Greek letters don't get mistranscribed as English words. Nabla versus delta versus gradient symbols are distinguished. Matrix notation is preserved. When the professor says 'del cross B equals mu nought j,' we give you ∇×B = μ₀J in LaTeX, not a sentence that reads like a bad translation.
For problem sets, Distill generates variations on the problems your professor worked through in class. Same structure, different numbers. Great for drilling the technique without memorizing the one example you saw.
Real workflows. No fluff.
Distill transcribes spoken equations directly into LaTeX. Maxwell's equations, Navier-Stokes, Laplace transforms — all preserved with Greek letters, subscripts, and operators intact. No more retyping.
Snap a photo of the whiteboard at the end of class and upload it with the lecture audio. Distill aligns written work to spoken explanation — you see the equation and hear what the professor said about it at the same timestamp.
Distill builds a knowledge graph linking concepts across lectures. 'Convolution' in Lec 4 links to 'impulse response' in Lec 7 and 'transfer function' in Lec 11. You see the skeleton of the course visually.
Upload a completed problem and Distill generates 5–10 variants with different numeric values and slightly different constraints. Great for quals prep or when your prof didn't give you enough practice.
When your prof skips three algebraic steps, Distill fills them in as an expandable 'show work' section. You can verify your own intuition or catch where you lost the thread.
Stars mark the top 3 features for engineering students.
| Feature | Distill | Generic AI tools |
|---|---|---|
| ★LaTeX equation output | Native LaTeX with Greek, subscripts, operators | Plain text, mangled |
| ★Whiteboard photo alignment | Audio ↔ image sync | Not supported |
| ★Concept graph generation | Cross-lecture knowledge graph | Per-session bullet list |
| Problem set auto-generation | Numeric variants built in | Not supported |
| Derivation step expansion | Filled-in algebra | Generic summary |
| Unit/dimensional analysis | Preserved and checked | Often dropped |
| Code block handling (for CS lectures) | Syntax highlighting | Plain text |
| Overleaf export | One-click .tex bundle | Manual |
| Long sessions (lab + lecture) | Up to 15 hrs | 30 min |
| Greek letter recognition | 99.1% accuracy | ~82% |
“LaTeX was the killer feature. I used to retype every equation from lecture into my notes. Now I record, upload, and have Overleaf-ready .tex files in 15 minutes.”
Hassan A.
EE senior, quals bound
“The whiteboard alignment is wild. I take a photo at the end of class, upload the audio, and the board writing lines up with exactly what the prof said at that moment. That's never existed before.”
Zoe L.
ME junior, FSAE captain
“Problem set variants saved me for thermodynamics. The prof gave us three examples and expected us to nail a 15-question exam. Distill made 30 more I could drill.”
Marcus J.
ChemE, headed to grad school
ChatGPT and generic transcribers miss the details that matter. Distill was built for your workflow.
Generic transcribers give you 'mu sub zero times j' as literal text. Distill gives you μ₀J in LaTeX. Your Overleaf editor renders it correctly on the first try.
We're the only transcription tool that aligns uploaded whiteboard photos with lecture audio. If your prof writes more than they say, you need this.
Engineering courses are dependency DAGs. You can't learn Laplace transforms without convolution, can't learn convolution without impulse response. Distill surfaces those edges so you know what prerequisites you're missing.
Pattern-matching to solved examples only works on the easiest exams. Distill generates numeric variants so you train the technique, not the numbers.
Priya had six months to prep for EE quals covering signals, E&M, and solid-state. Her undergrad notes were fine but scattered across 400+ iPad PDFs. She needed to consolidate.
She uploaded every lecture recording from her junior and senior years to Distill — 280 hours of audio. Distill built a unified concept graph linking signals/systems to probability to stochastic processes. It generated 400+ problem variants across topics.
During the six-month prep, Priya drilled problems by topic using the generated variants. She used the concept graph to identify her weak nodes — it showed solid-state as a thin cluster compared to signals — and prioritized there.
She passed quals on first attempt, top third of cohort. She keeps her Distill workspace live in grad school as a searchable reference for when she's revisiting an old technique.
Outcome: Passed PhD quals on first attempt with 280 hrs of lectures organized into a searchable concept graph and 400 drill-ready problem variants.
Plus unlocks unlimited capture, LaTeX export, and concept graphs. Pro adds 15-hour sessions for lab days, problem-set auto-generation, and Overleaf integration.
The problem set is due Tuesday. The professor's whiteboard is gone. Your audio is a blurry ghost of the real lecture. Distill was built for exactly this problem — let the machine do the transcription so you can do the engineering.