Introducing Yomite — an offline Japanese TTS reader with 14 voices

Today we’re launching Yomite for iOS, iPadOS, and macOS. It’s a Japanese text-to-speech reader that runs 14 character voices entirely on Apple Silicon devices — no cloud, no API keys, no upload.

Why we built it

Reading time keeps shrinking. We wanted to give people a different angle of access to books — by ear, on the move, during chores, while resting tired eyes.

Audiobooks are great, but their catalogues are limited. Meanwhile, Aozora Bunko hosts thousands of Japanese literary classics for free, EPUB stores carry millions more, and the web publishes endless long-form articles every day. If listening to all of that worked well, reading time would dramatically expand.

To make this work, we needed two things: a natural Japanese voice, and a privacy model that doesn’t ship text to a server. So we built both.

What it does

Read EPUB, Aozora Bunko, and web articles aloud. Drop in a local EPUB for chapter-by-chapter playback. Browse Aozora Bunko classics from inside the app. Paste a URL — the built-in article extractor pulls the body text only.

14 character voices. Cool 凛音エル, friendly 桜音, whispering 天深シノ, baritone 阿井田 茂, plus 10 more. Several voices ship with up to 7 emotion styles (normal, sad, whisper, shout, etc).

Fully on-device. After the initial voice download, your text never leaves your device. Yomite runs 音声合成 through オンデバイス処理 + オンデバイス処理, executing every inference locally.

CarPlay and Live Activity. Driving, doing chores, drifting to sleep — control playback from the lock screen, widgets, and CarPlay display.

Why offline

Cloud TTS comes with costs, latency, and privacy compromises. We chose on-device inference to dodge all three.

• Text stays on your device. Reading history is yours.
• Works offline. Subway, airplane, dead-zone — no playback interruption.
• Free with no metered cost. No cloud bill means we can keep v1 entirely free.

System requirements

• iPhone (A12 Bionic or newer) — iPhone XS / XR / SE2 onward
• iPad (A12 or newer) — iPad mini 5, iPad 8, iPad Air 3 onward
• Mac (M1 or newer)
• iOS 17 / iPadOS 17 / macOS 14 or later

Older A11-and-below devices can’t reach real-time inference speeds for v1.

What’s next

v1.0.0 is launch beta. We’re already planning:

• m4a / WAV export for narration creators
• Android version — timeline under discussion
• Additional voices — including possible collaborations with voice actors and ボイスモデル creators
• Chapter-level audio caching to make offline listening even smoother

Feedback and feature requests: support@yomite.app.

Happy listening.

— The Yomite Team