How Big Are Yomite's Voices? — A Storage Guide to Voice Models

Ever downloaded a few Yomite voices and found your device’s free space shrinking faster than expected? Because Yomite’s speech synthesis runs entirely on your device, it works differently from cloud TTS apps: the voices themselves need to live on your device, not on a server. Here’s what makes up that storage footprint, and how to download smartly.

Each voice is 127-270MB

A single Yomite voice model runs 127-270MB. That range covers both the free tier’s two voices (Mai and fumifumi) and the 30+ voices unlocked with Premium. Exact size varies a bit by voice and how many speaking styles it has, so downloading several voices you like can easily add up to several hundred megabytes, or more than a gigabyte.

Regular quality vs. high quality — which to pick

For each voice, Yomite lets you choose between regular quality and a high quality version. The high-quality version sounds closer to natural speech but takes up more space; regular quality is more compact.

• Mostly listening on the go — commuting, doing chores — regular quality is plenty
• Listening closely through a Bluetooth speaker, or you just care about audio quality — high quality is worth the extra space

If storage is tight, a good approach is to try a few voices at regular quality first, then upgrade only your favorites to high quality.

The shared language processing model (667MB) downloads once

There’s also a language processing model behind every voice — 667MB — that handles text analysis. It isn’t downloaded per voice; it’s downloaded once and shared across all voices. That means the very first voice you download brings in this 667MB alongside its own file, making it the single biggest download. Every voice after that only adds its own 127-270MB.

For example, using both free-tier voices (Mai and fumifumi) at regular quality means roughly the 667MB shared model plus the two voice files — somewhere around 900MB-1.2GB total. If you’re on Premium and decide to download every one of the 30+ voices, the voice files alone could add up to several gigabytes. You don’t need every voice sitting on your device at once, so it’s more practical to keep only the ones you actually listen to.

If playback stutters or stalls, check your free space first

As noted in the Web Reader troubleshooting guide, stuttering or stalled playback is often a storage issue. As a rule of thumb, keep roughly 1GB of free space available for the language model plus your voices. If you’re planning to queue up a long novel or several web articles, factor that headroom in alongside whatever voices you’ve downloaded.

A note on supported devices

Yomite’s speech synthesis is built around Apple Silicon’s processing power, and requires iPhone (A12 Bionic or later), iPad (A12 or later), or Mac (M1 or later) running iOS 17, iPadOS 17, or macOS 14 or newer. That’s a separate question from storage, but on older devices, inference speed won’t be practical — worth checking your device is supported before you start downloading voices.

The takeaway: the first voice is the heaviest download

Downloading Yomite voices works like this: the first voice you download is the biggest download (667MB shared language model plus 127-270MB for the voice itself), and every voice after that only adds its own file. Sticking to the two free voices keeps you in the hundreds-of-megabytes range; if you’re on Premium and want to explore more voices, starting with regular quality and upgrading only your favorites is the storage-friendly way to go.

Knowing this ahead of time means no surprises when you check your device’s storage after a download spree.

---

Get Yomite on the App Store →

Related: Why Yomite’s Speech Synthesis Runs Entirely On-Device · Web Articles, Read Aloud — The Web Reader Tutorial

Questions or feedback: [email protected]

— Yomite Team