PrivateLocalOpen source

Private by default

Your documents and generated audio stay on your device by default, with no accounts, tracking, or telemetry.

Local first

OpenReader is designed to keep reading useful offline, with local voices and model files you control.

Open source

The app is built in the open so people can inspect it, improve it, and trust how it works.

Read anything.Listen when you want.

Bring PDFs, EPUBs, text, articles, and feeds into one focused reader. Read silently or listen with voices running on your device.Bring PDFs, EPUBs, text, web articles, RSS feeds, and public catalogs into one focused reader. Read silently or listen with built-in and open voice models running on your device.Read PDFs, EPUBs, articles, and feeds silently or with voices running on your device.

Web

Library

1h 34m42%

5h 38m36%

9h 12m68%

PDF

Attention Is All You Need

58 min5%

15 min

Add a reading

Paste, drop, or upload

Private

No account. No telemetry.

Books, PDFs, feeds, voice choices, and generated audio stay on your device by default.

Sources

One place for every source.

Open PDF, EPUB, text, Markdown, and article URLs, or browse RSS feeds and OPDS catalogs.

Articles

Keep the shape of the story.

Readable articles preserve headings, lists, links, and images, with a safe path back to blocked originals.

Your choice

Read or listen.

Choose No narration for a focused reader, a built-in system voice, or a downloadable local model.

Sync and progress

Stay with every word.

Follow reading keeps text, pages, the seek bar, and spoken words together. Your position and generated audio progress stay with the document.

Appearance

Native by default. Yours by choice.

Follow your device in light or dark mode, then tune the reading view and theme when you want.

Narration is optional

Choose how reading sounds.

Use No narration, a voice already built into your device, or an open model downloaded once for private local speech.

No narrationUse the complete reader without generating audio.

Apple System Voicebuilt in, instant startup

Kokoro 82Mfast, lightweight narration

Qwen3 TTS 1.7Bexpressive multilingual voices

Voxtral 4B18 voices across 9 languages

Higgs Audio v2voice cloning from a short sample

Real benchmarks

Local voices are fast enough to feel immediate.

Measured on real hardware with the local TTS path. Higher RTF means faster generation.

Fastest run

23.0 min

audio / compute min

Kokoro on Apple M2 Pro / 32 GB.

Chip	Model	TTFATime to first audio before playback can start.	App RTFReal-time factor through the full desktop backend path with word-sync alignment.	Pure RTFRaw model real-time factor without app or aligner overhead.
Apple M2 Pro / 32 GB	Kokoro	0.28s	23.0x	22.2x
Apple M2 Pro / 32 GB	Qwen3	4.41s	1.5x	1.6x
Apple M2 Pro / 32 GB	Higgs	6.54s	1.0x	1.0x
Apple M2 Pro / 32 GB	Voxtral	16.22s	0.4x	0.4x