Olzaras: AI Voice PDF Reader

Offline reader · 31 languages · Android · Coming soon

What it does

Olzaras reads books and documents aloud. You open a PDF, an EPUB, a Word file or plain text, pick a language and a voice, and the phone reads it to you with a natural neural voice, in the car, on a walk or with the screen off. While it reads, the text on screen follows the voice word by word, with the current sentence highlighted, and everything happens on the phone itself.

No internet

The app does not have the Android internet permission. It cannot use Wi‑Fi, 4G, 5G or any other connection, because the system never grants it one. It works in airplane mode, and it never needs a connection again after it is installed. The voices are installed together with the app from Google Play; nothing is downloaded later, nothing is sent anywhere.

It also asks for no permissions: your files are opened through Android’s own file picker, so the app has no permission to read your storage, and the playback notification needs none.

What it reads

PDF (text extracted on the device)
EPUB, DOCX (Word), FB2, HTML
TXT, Markdown and any plain text, or text you paste or type directly
Up to 600,000 characters per document (a long novel)

31 languages, 10 voices

English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Russian, Ukrainian, Bulgarian, Czech, Slovak, Slovenian, Croatian, Romanian, Hungarian, Greek, Turkish, Swedish, Danish, Finnish, Estonian, Latvian, Lithuanian, Indonesian, Vietnamese, Hindi, Arabic, Japanese and Korean.

The same ten voices (five female, five male) speak all 31 languages. They come from the Supertonic 3 model (MIT-licensed code, OpenRAIL-M voice weights) and run entirely on the phone. You can play a sample of each voice before choosing it, and give the voices your own names.

The app’s own interface is translated into all 31 languages, and the menu language is chosen separately from the language being read.

The player

Play / pause, skip 10 seconds back or forward; press and hold Play to stop, with a choice between starting over and continuing from where you were
Five reading speeds: Very slow, Slow, Medium, Fast, Maximum (0.8× to 1.3×)
Switch the voice, the speed or a narrator setting while it reads: the new voice picks up from the same word, without starting over
Elapsed, remaining and total time for the whole document, estimated from the text and refined as it reads
Sleep timer: after 5, 15, 30 or 60 minutes, at the end of the paragraph, or at the end of the chapter
Keeps reading in the background, with controls in the notification, on the lock screen and on headset or car buttons
Remembers where you stopped in the document you are reading, so the next Play continues from there
Full-screen reading view with your choice of font (Sans, Serif, Mono or the system font), text size, text colour, highlight colour and background
Smart narrator: skips page numbers, running headers, Project Gutenberg boilerplate, footnote marks, scene breaks, hyphenation, duplicated lines, tables of contents and speaker labels; each of the nine can be switched off
Four reading backgrounds (Light, Dark, Dimmed, Snow) and sixteen colour themes
Labels for screen readers (TalkBack) on every control

Record to MP3

Press REC and the reading is saved as an MP3 file (320 kbps) in the phone’s Music/Olzaras folder, so you can listen to it in any player or keep it. The recordings are also listed in the app’s Library, where you can play, share, rename and delete them.

Every recording carries a tag marking it as AI-generated speech and, when the phone has enough memory for it, an inaudible watermark as well, which is the disclosure the voice model’s licence requires.

Good to know

On the first launch the app prepares a working copy of the voices (about 500 MB), which happens once.
Speech starts a few seconds after Play: the first sentence is being synthesized, and the screen shows the countdown. Reading needs about 500 MB of free storage to start.
The MP3 is prepared after you stop the recording, not while it plays; on slower phones this takes roughly as long as the recording itself. Until the file is ready the Library shows “Preparing…” and the REC button is disabled.
Changing the speed or the voice during a recording finishes that recording and keeps the file; press REC again to record at the new setting.
On slower phones, skipping far ahead can pause the voice for a couple of seconds while the next passage is synthesized, after which it resumes.
A pause longer than 10 minutes ends the reading session and removes the notification, so that the app does not keep the phone awake; your place is saved and Play continues from there.
The speech is synthesized continuously on the processor, so reading uses more battery than playing a recorded audiobook: on a mid-range phone, six hours of continuous reading used about half of the battery.

Privacy

The app has no account, sign-in, advertising, analytics, crash reporting or tracking, and your books and recordings stay on your phone. Details: privacy policy.

Price

One purchase on Google Play, with no subscription, in-app purchases or separate “pro” version. Refunds: see the refunds page.

Requirements

Android 10 or newer
A phone with 4 GB of RAM or more; low-memory (Android Go) devices are not supported
OpenGL ES 3.0 (any phone from roughly the last decade)
About 1.1 GB of storage in total: the app with its voices (about 560 MB, installed once from Google Play, nothing downloaded later) plus the working copy of the voices that the app prepares on first launch (about 500 MB)

Support and legal

Developer: Diyan Hristov, trading as DH Studio Technologies
Support: support@dhstudio.dev
Legal: Privacy policy · Terms of use · Refunds