Accessibility & productivity · Available on the App Store

Vox

Turn PDFs, pasted text, and web articles into calm, controllable audio.

Vox is a focused read-aloud library for iPhone, iPad, and Mac with device voices, saved audio, and a deliberately separated research path for offline and cloud narration.

Explore the case study

App Store

Category: Accessibility & productivity
Status: Available on the App Store
Platforms: iPhone · iPad · macOS
Built with: SwiftUI · AVSpeechSynthesizer · PDFKit

The product problem

Why Vox exists.

Long documents are often trapped in a visual reading context. Vox creates a listening workflow without turning document import, voice choice, and audio reuse into separate utilities.

What the product does

The useful parts, stated plainly.

These capabilities are grounded in the project’s README, metadata, source structure, and checked-in product documentation.

Import the words

Open PDF, RTF, HTML, and text documents; paste or write text; or extract the readable content from a web URL.

Choose the right voice

Use system voices for a local baseline and compare other clearly labeled voice paths when available.

Control the reading

Keep play, pause, voice choice, sentence highlighting, and reading controls near the content.

Save reusable audio

Render a reading into an audio clip and return to it from the Audio library.

Organize a library

Use folders, recents, unfiled items, search, and grid or list views across different reading sources.

Read day or night

A shared design system supports light and dark presentation across the Apple device family.

Designed for

People with a specific job to do.

People who listen while commuting or doing hands-on work
Students and researchers working through long documents
Readers who benefit from synchronized audio and text
Anyone turning saved articles into a listening queue

Product choices

What makes it distinctly Orion.

The free baseline uses voices already available through Apple’s speech APIs.
Document storage remains local, with optional private iCloud sync through the user’s Apple account.
Voice quality, privacy, licensing, and cost are treated as separate product decisions rather than one “AI voice” label.

Product gallery

The real interface, inside the right frame.

Every image below is a clean application capture from the product repository. Orion supplies the unbranded hardware frame.

Vox reading library with folders and recent documents on iPhone — 01Folders and recent readings make audio a repeatable habit.

Vox import options for text, documents, and web pages on iPhone — 02Text, documents, and web pages begin in one import flow.

Vox document reader with playback and audio controls on iPhone — 03Playback, voice status, and saved audio stay beside the text.

Vox voice library on iPhone — 04Voice choices are grouped by how and where synthesis happens.

Vox research

The method and the limitation belong together.

Public-safe summaries from the project’s substantive research and product documentation.

June–July 2026

A speech backend is a product choice, not just a model choice

Question: Which text-to-speech paths balance quality, privacy, app size, licensing, latency, and predictable cost across Apple platforms?
Method: Vox evaluated Apple’s native speech stack, bundled ONNX approaches, and cloud synthesis behind one app-facing playback interface.
Key finding: Apple system speech is the dependable baseline: it is available through AVFoundation, uses installed device voices, and avoids a model bundle or metered backend. Bundled neural speech can improve the voice palette but introduces runtime, size, performance, and licensing review.
Limitation: The evaluated Kokoro English path depends on espeak-ng in the tested configuration, creating unresolved GPL obligations for a closed-source paid release. Neural playback performance also varies by device.
Product implication: The product keeps system speech as the safe baseline and treats the bundled neural path as non-shippable until the licensing gate is resolved.

Public sourcesApple: AVSpeechSynthesizer ↗sherpa-onnx license ↗Kokoro model card ↗

July 5, 2026

Quote the cost before synthesis

Question: How can a cloud voice show a deterministic user price before any text leaves the device?
Method: The study compared character-priced and token-priced synthesis, then modeled credits, cache behavior, App Store commission, and abuse controls.
Key finding: Character-based pricing is easier to quote exactly. A content-and-voice hash can also make an already synthesized passage replay without a second charge.
Limitation: A client-side credit ledger or direct provider credentials are not sufficient for a production economy. Usage, receipts, and spending limits must be enforced server-side.
Product implication: Cloud synthesis remains behind explicit confirmation, with a server-side proxy, receipt validation, and spending limits identified as release gates.

From product story to production

Want to build something as considered as Vox?

Bring us the early idea, the difficult workflow, or the product that needs a clearer next chapter.

Strategy · Design · EngineeringDiscuss your product

Vox