Speech Central Pack: Kokoro

Enjoy natural English Kokoro voices in your Android reading apps—privately and offline.

SC Kokoro is a system text-to-speech engine designed for books, articles, and other long reads. On compatible phones, it uses the GPU for the demanding work of generating speech. Compared with Kokoro engines that rely on the CPU alone, this gives capable devices more room to prepare upcoming speech and helps reading feel more responsive. It significantly reduces heat generation, which in turn leads to lower battery consumption.

Choose from 28 English voices, including US and UK options. Download the voice data once, then generate speech on your device. The text you listen to is not sent to a cloud speech service.

SC Kokoro works with apps that support Android text-to-speech engines. You can select it in a reading app that offers a TTS engine choice. For the smoothest long-form listening, we recommend Speech Central: its advanced buffering can prepare upcoming passages and reduce interruptions. Playback in other apps depends partly on how they request and buffer speech.

Speech generation speed depends on your device. With effective buffering, long-form reading can work well even on mid-range devices. On slower devices, SC Kokoro can still generate high-quality audio files through apps that support audio export, although the process may take longer. Tasks requiring immediate spoken feedback, such as TalkBack, are much more demanding and may be challenging even on high-end devices.

The app checks compatibility before downloading the model and shows an estimate of the experience you can expect. The maximal performance is achieved when the screen is off while reading.

Source code and licenses: https://github.com/speech-central/sc-voices-android