Piper TTS in WithAudio: German First, Many Languages Ahead

Piper's simple architecture and reasonable performance on local machines has always made it a very attractive option for being used along WithAudio's no subscription, local and private mindset.

So far, most new languages and voices in WithAudio were added by extending the built-in TTS engines that run inside the app (the same paths that power voices you already use). That is a good fit when we control the full stack, but Piper alone ships a large catalog of languages and downloadable voices. Reimplementing every one of those models inside our in-app engines would be slow and brittle.

Starting in desktop v0.1.75, Mac and Windows both integrate Piper as a first-class path for text-to-speech. WithAudio manages the Piper process for you. It runs as a separate OS process, but the app starts it, keeps it in step with playback, and talks to it locally so audio and text highlighting stay in sync. You keep the same reading workflow you already know. All you need to do is to make sure you are running the latset version of WithAudio.

What is Piper?

Piper is a fast, local neural text-to-speech system: you send text, you get audio on your machine, without routing speech through a cloud API. It grew out of the Rhasspy Piper project (that repository is now archived); active development continues under the Open Home Foundation as piper1-gpl, with pip install piper-tts, command-line and server modes, and a wide set of pre-trained voice models maintained by the community. For readers, the practical upshot is simple: high-quality offline speech, a large catalog of languages, and a path to add voices that were not wired into WithAudio before. WithAudio is using the Rhasspy Piper.

German today: Thorsten-Voice

For this release we started with German, using the Thorsten-Voice models associated with Thorsten Müller’s project. A long-running, openly shared effort to make free, offline, high-quality German TTS available without license friction. If you read or study in German, you can pick a Thorsten-backed Piper voice in the app and hear narration that matches the language of your text instead of stretching an English voice across umlauts and grammar.

You can browse German voices on the site to hear samples before you import a long document. As with our other engines, use Using voices and languages for filters, downloads, and quick access. If you do not see Piper yet, update to v0.1.75 or later and check the voice list and any one-time setup steps for the Piper engine.

Where this goes next

Piper is not a one-off voice pack, it is a platform. Shipping it on both Mac and Windows in v0.1.75 is the foundation for many more languages and voices on with.audio without each one requiring a bespoke integration inside the core app. German is the first cut; the list will grow.

Try it

WithAudio is a one-time purchase desktop reader with a free trial and a seven-day money-back guarantee. Update to v0.1.75, add a Piper voice from the picker, and press play on your next German document or keep an eye on the changelog as we widen the catalog.