Upload any PDF · Listen anywhere

Listen your way
through any book.

Upload a PDF and get every page distilled into a crisp 100-word summary — read it or listen to it, with sentence-level audio sync.

Open library → See how it works
Learn mode

Listen while you commute, read when you focus

Every page of your PDF becomes a tight ~100-word summary, ready to play. Sentences highlight as they're spoken — tap any line to jump straight to it.

🎧
Continuous audio playback Pages auto-advance so you can listen through an entire chapter hands-free.
✦
Sentence-level sync Each sentence highlights as it plays. Tap any sentence to seek directly to it.
⚡
Chapter navigation Jump to any section from the sidebar. All 10 chapters, always a tap away.
Chapter 2 · page 74
Post-training aligns model outputs with human preferences, moving beyond raw capability.
Sampling techniques — often overlooked — significantly shape what the model actually says without touching its weights.
Temperature controls randomness; top-p and top-k narrow the candidate token pool before sampling.
Instruction fine-tuning teaches the model to follow directives rather than just continue text.
⏸
0:18 / 0:42
Review mode

Chapter summaries, section by section

After listening, switch to Review to scan each chapter as structured bullet-point summaries — one card per section, collapsed by default so you can skim or dive in.

≡
Bullet-point summaries Each section distilled to its core ideas. Expand any card to read the full summary.
⊟
Scan the whole chapter fast All sections collapsed by default — scroll to see what you remember, expand what you don't.
✦
Visual recall cards AI-generated concept images and flip cards available in Pro.
Review · Chapter 2
Post-Training ▾
Aligns outputs with human preferences
RLHF uses reward models to guide updates
SFT is the foundation before preference tuning
Sampling Strategies ▸
Fine-Tuning ▸
How it works

Works with any PDF

Upload once — the pipeline processes your book and streams pages as they're ready.

1

PDF → Page summaries

Each page is condensed to ~100 words using a local LLM, preserving technical accuracy while making it listenable.

2

Summaries → Audio

A text-to-speech model converts every summary to natural MP3 audio with sentence-level timestamps for playback sync.

3

Chapters → Summaries

Each chapter is broken into sections with bullet-point summaries — ready to review after listening.

Upload your first book

Drop in any PDF — summaries and audio are ready in minutes.

Open library →