Algebra II · Unit 4
Eigenvalues and eigenvectors
λ is an eigenvalue of A if A·v = λ·v with v ≠ 0.
det(A − λI) = 0
λ₁ = 5 · λ₂ = 2
4123
Filmina explains your class PDFs out loud: studying at home feels like being back in the lecture, minus the professor's anecdotes.
A ninety-slide deck doesn't get studied: it gets abandoned around slide twenty. In class it made sense because someone was explaining it; alone, with the same slide in front of you, it doesn't.
Reading in silence you skip half a paragraph without noticing and have to read it again. Listening to the lecture while you look at it, the session moves on its own: you finish sooner, and the next day you remember what you heard.
The model that explains and the voice that speaks both run on your processor. Nobody on the other side sees your notes, because there is no other side.
You download it and open it. No sign-up, no verification email, nothing to recover when you forget the password.
Since there is no server to keep running, there is nothing to charge you for every month. Filmina is paid for once.
You set it to summary and hear the whole document in 19 minutes instead of 58, with the three hard ones already marked.
On full, the frame follows the voice: half an hour later you are still there, not on your phone. And when it ends, it asks.
You open the document the night before and it picks up where you left it. The one you never understood you put on extended.
The slide is split into segments of one or two ideas, and then it is verified in code that not a single block of text was left unexplained. Coverage doesn't depend on the model behaving: it is checked.
The frame surrounds the section it is talking about and slides to the next one. No fill, no veil: any layer over the page, however soft, gets in the way exactly where you need to read.
P(A|B) is said «P of A given B», not «p bracket a bar b». The notation is translated before it reaches the voice, which is where text-to-speech readers fall apart.
The preamble and the wrap-up are stripped out («we're on slide 22 of 62», «let's take a look», «in short») because they are the seconds you spend waiting for the part that matters.
It breathes and pauses where it should.
Kokoro 82M local voice model
«An architectural pattern is a reusable solution to a problem that shows up again and again in one and the same context.»
Four options, and the wrong ones are the usual confusions, not absurdities.
Writing the word is quite a bit more than recognising it among four.
You talk for thirty seconds and it tells you which ideas you covered.
Answered from memory, in words. Whether you had it is yours to say.
Free
The first 10 documents
US$ 39
Once, forever
US$ 78
−50 %
US$ 19
With your university email
US$ 38
−50 %
Only to download and activate it. After that, no: the explanations and the voice are generated on your machine.
Any Mac with Apple Silicon (M1 onwards) and macOS 13 or later. The model takes up around 10 GB on disk the first time.
If the page is a photo with no text layer, Filmina tells you instead of making things up. You have to run OCR on it first.
It marks them as a figure and tells you to look. The model reads text: it would rather tell you there is something there than explain a chart it never saw.
On summary and full it stays with what the slide says, and it is verified in code that every block was covered. Extended is the only level that adds outside context, and it says so.
A text-to-speech reader tells you what is written on the slide; NotebookLM turns the whole PDF into a podcast. Filmina explains each slide while you look at it and points at the block it is talking about. Compared in detail: against Speechify, against NotebookLM and against ChatGPT.
In English or in Spanish, whichever you pick, whatever language the PDF is written in.
Yes. One licence, every update to version 1 included, and the app keeps working even if you never pay again. If a big version 2 ever arrives, it will be an optional upgrade.
Drop in that PDF you've been putting off and hear it explained. Thirty seconds is enough to know whether it's for you.
No account · macOS 13 or later
by @ignaprados