Timbra turns the documents you never find time to read — two-column papers, scanned books, textbooks full of tables — into audio you'll actually finish: properly chaptered, with a reading page that follows every word. Hear your first paragraph free, then see the measured quality before you pay a cent.
Timbra routes each document to the parser that handles it best, then tells you what it found before it starts reading.
Drop in a week of preprints. Get them chaptered by section, with the maths spoken properly and the bibliography left where it belongs. Ask the document a question and it answers with the page number.
340-page thesis → 11 h 20 m · ready in 14 min
Chapter checkpoints, per-chapter summaries and a quiz built from what you actually listened to. Highlights and voice notes export straight to your notes app.
Average session 34 min · 2.1× replay on hard chapters
Word-level highlighting locked to the audio, Atkinson Hyperlegible and OpenDyslexic, adjustable spacing, full VoiceOver and TalkBack support. Not a mode — the default engineering standard.
Sync accuracy < 100 ms · WCAG 2.2 AA
Translate a paper into Portuguese and hear it at 0.8×, with the original text beside it. Tap any word for a definition without losing your place.
6 launch languages · 23 more in beta
Wind-down slows the narration, brings up the rain and fades out on a timer. When you wake up, it resumes thirty seconds before you stopped paying attention.
Sleep timer with rollback · screen never whites out
Shared shelves, seat management, and a per-document audit of what was read aloud and what was skipped. Data stays in the EU.
EU regions only · DPA + sub-processor list
We pull two passages out of your document — one plain narrative, one thick with names and numbers — because that second one is where cheap narration falls apart. Everything you change updates the price as you go.
The harbour kept its own hours. Long after the town had folded itself away for the night, the cranes went on turning, and the light on the far mole answered them once every four seconds, patient as a metronome.
Follow along word by word, or turn the text off entirely and just listen. Your place syncs across every device, down to the sentence.
Every chapter is transcribed back and compared to the source, scored for naturalness, and checked for voice drift. You see the numbers, what pulled them down, and exactly what it would cost to fix.
Every number here is measured on your file, not quoted from a benchmark. We publish the method, the model versions and the seeds behind each render.
Read the measurement methodA soundbed sits under the narration at a level you set, and ducks itself when the voice comes in. Every thirty days the Season changes — a new set of beds, a new palette, a new nightly piece.
Tonight: Rain on the mole
Generated fresh at your local dusk, seeded from the date so everyone on Season 03 hears the same twelve minutes tonight. Tomorrow it's gone.
Hearth — low strings, room tone, a fire that never quite settles. Chosen for you: it's 23:10, it's raining outside, and you've been on the same chapter for three nights.
Nine of twelve. Finish the Season and it stays in your library — the only way to keep a night once it's passed.
The beds are generated, not licensed. Procedural pads and field textures, composed on your device from a daily seed. No royalties, no repetition, no track that a rights-holder can pull out from under you.
You can tell people what you're reading and let them hold you to it. You cannot scroll here, and nobody can see how many people are watching you.
Nine people are reading with the same soundbed right now. No names, no chat, no leaderboard — just the knowledge that the room isn't empty. Turn it off and it never comes back.
Pick a book, set chapter checkpoints, and leave thirty-second voice notes in the margin for each other. Everyone hears them at the exact moment they reach that paragraph.
Share cards carry your progress and your own words — never the audio, never the book's text. What you upload stays yours and stays private.
Per-document prices are calculated from the actual work your file needs — pages, scan quality, language, voice — and then fixed. Subscriptions come with an allowance in hours, so you always know where you stand.
€0
€7.99 / MONTH
€14.99 / MONTH
from €4.90 / BOOK
Upload it. Hear a paragraph in four voices. Decide after that.