You queue up a podcast on macro you've been putting off. Three minutes in, Wisot pauses: "What's the difference between core and headline inflation?" You whisper the answer — the guy across the aisle doesn't notice. Forty minutes later, you walk into work and the chart on your screen actually makes sense.
Built at MIPT · Backed by Bortnik Foundation
You listen to everything. You remember almost none of it.
Wisot listens to any audio on your phone and pauses at the right moment. Asks you a question. You answer by voice. Playback resumes.
14-day Pro free · No card · 3 free hours/month forever after trial
WISOT
00:00
LIVE SESSION
Session in progress
Start listening
…the function is differentiable at a point if there exists a finite derivative…
TUTOR AI
Ask a question…
MIPTFounded by students
Sechenov UniversityCognitive science
Bortnik FoundationStudent Startup Grant
Peer-reviewed researchRoediger, Karpicke, Bjork
Wisot LLCRussian legal entity
COST OF PASSIVE LISTENING
Two hours after a lecture, you remember a fifth.
Ebbinghaus showed it in 1885. Nothing about your brain has changed since — the things you don't use, you lose. Podcasts, lectures, audiobooks pass through you like water through a sieve. You spend the hours. You keep a sentence.
Enough. Send me the APK →~20%
what you'll recall from a lecture two hours later, with no review
0×
SOURCE: Ebbinghaus (1885) · Roediger & Karpicke (2006)times most people revisit a podcast they "loved"
How Wisot fits into your audio
01
Play anything
Start a lecture on YouTube, a podcast in Spotify, an audiobook in Audible. Wisot listens at the system level — the source doesn't matter.
01
01
Play anything
Start a lecture on YouTube, a podcast in Spotify, an audiobook in Audible. Wisot listens at the system level — the source doesn't matter.
02
02
Recognition runs on-device
Speech is transcribed locally via sherpa-onnx. Your raw audio is never uploaded to our servers. Recognition runs without a network round-trip — no lag.
02
Recognition runs on-device
Speech is transcribed locally via sherpa-onnx. Your raw audio is never uploaded to our servers. Recognition runs without a network round-trip — no lag.
03
It pauses at the right moment
Wisot catches the key idea — a definition, a number, an argument — and pauses playback. Asks a real question.
03
03
It pauses at the right moment
Wisot catches the key idea — a definition, a number, an argument — and pauses playback. Asks a real question.
04
04
Answer by voice — and continue
Speak the answer or type it. The AI grades meaning, not exact words. Playback resumes. The memory locks in.
04
Answer by voice — and continue
Speak the answer or type it. The AI grades meaning, not exact words. Playback resumes. The memory locks in.
Testing beats re-reading
56% vs 42%
Roediger & Karpicke (2006), Psychological ScienceThe group quizzed after a passage remembered it significantly better a week later than the group that just re-read it. Re-reading feels like learning. It isn't. Wisot replaces re-reading with a small test at the right moment.
"High utility"
Dunlosky et al. (2013), Psychological Science in the Public InterestA meta-analysis of ten study techniques. Practice testing was the only one to earn the top rating. Highlighting and underlining earned the lowest. Wisot embeds the best technique inside the audio you'd play anyway — and removes the friction.
INSIDE WISOT · v0.8.1
This is the real app, not a render
Screenshots straight from v0.8.1 — the build you'll get in your inbox. No retouching, no design comps. Just what actually runs on your Android.
SHOT ON PIXEL 7 · v0.8.1 STABLE · APRIL 7, 2026
Four things no one else does
Answer by voice — hands free
Running, cooking, driving — Wisot asks the question and you say the answer out loud. The AI grades meaning, not exact words. Playback continues on its own.
Works with any app
YouTube, Spotify, Audible, Telegram lectures, local mp3 — Wisot listens at the system level. No uploads. It just works with whatever you're playing.
It adapts to you
Under the hood: Bayesian Knowledge Tracing, Bloom's Taxonomy, FSRS-5 spaced repetition. When you're solid, the questions get harder. When you're lost, Wisot backs off.
Speech recognition runs on-device
ASR runs locally via sherpa-onnx. Your raw audio is never uploaded to our servers — not for "quality", not for analytics. Transcripts and learning progress sync across devices through your Google account.
USE CASES
Where Wisot fits into your day
You're making pasta. A stats podcast plays through the speaker. Stirring sauce, Wisot asks: "How is p-value different from effect size?" You answer out loud over the pan. No hands, no screen. By the time the pasta's done, three ideas have stuck.
7 a.m., Kahneman audiobook, kilometer six. Wisot catches the moment after a key claim about System 1 and System 2 and pauses. You answer without breaking stride. You get home and that chapter doesn't dissolve by evening — for once.
WHAT NO ONE ELSE DOES
Wisot is the only player that does all of this at once
Transcription tools give you text. Flashcard apps need manual work. Podcast apps just play. Wisot does the thing none of them do.
01
The only tool that pauses any audio in real time
02
The only one that works with system audio — no uploads, no transcripts
03
The only one with on-device speech recognition — your raw audio never leaves the phone
04
The only one with a knowledge model that adapts as you do
PRICING
Pay less per month than a paperback
14-day Pro free. No card to install. After the trial — 3 free hours a month forever, or one of the plans below.
14 days of full Pro — free. No card. No auto-charges.
Starter
for casual listeners
$5/ mo
- 15 audio hours per month
- Voice answers, graded by meaning
- Basic adaptive difficulty (BKT)
- Markdown export
Billed monthly · Cancel anytime
Most popular
Active
for students and daily learners
$10/ mo
- 60 audio hours per month
- Full adaptive engine: BKT + Bloom + FSRS-5
- Visual context: Wisot sees your screen
- Export to Anki, Notion, Obsidian
- Priority support
Billed monthly · Cancel anytime
Mastery
for people who live in audio — researchers, analysts, clinicians
$25/ mo
- Unlimited hours
- Everything in Active
- Multiple knowledge libraries
- Early access to new models
- Direct line to the team
Billed monthly · Cancel anytime
After the trial — 3 audio hours per month, free, forever. No card, no nags.
THE TEAM
Three people. Backed by a grant.
Our goal is to make quality personalized learning accessible to everyone.
Mikhail Orlov
CEO, co-founder
Yulia Galkina
CTO, co-founder
Vladimir Antonets
Mentor, research advisor
Built with support from the Foundation for Assistance to Small Innovative EnterprisesFrequently asked questions
What we shipped this week
SHIPPED
On-device speech recognition — GigaAM v3
44100Hz → 16000Hz resample, recognition runs entirely on-device.
RESEARCH
FSRS-5 spaced repetition scheduling
Integrated FSRS-5 algorithm based on Ye et al. (2024).
DECISION
Voice + text instead of voice-only
Added text input: accessibility, noisy environments, user preferences.
OPEN TESTING
Join the Wisot open test
Wisot is an Android app for active learning from any audio. Download the APK with one click and start learning from any sound today.
Security verification
KasperskyCleanVerified by OpenTIPView report
VirusTotalReport available after release70+ antivirus engines
SHA256
A9D8221836FB7D09B897A15F50AED652B3876889EF660F09EAE5B2367B39D8BAStartrememberingthings.
14 days of Pro, free. No card. Android only.
Get the APK172 MB · Android 8+ · Bortnik-backed