Offline Voice Translators : the Gemma

The Gemma Translator is a handheld voice translation device that performs speech recognition, translation, and text-to-speech processing entirely offline. Built around a Raspberry Pi 5 with 8GB of RAM, the open-source device combines a 480×320 touchscreen, microphone, speaker, push-to-talk buttons, rotary encoder, and a 3D-printed enclosure. Google’s Gemma 4 E2B model runs locally through LiteRT-LM, while Moonshine handles speech recognition and Kokoro generates spoken translations, eliminating the need for cloud processing after the initial setup.

The device supports two conversation modes for either single-user or face-to-face translation, with physical controls replacing touchscreen navigation during conversations. Translation runs at approximately six tokens per second while using less than 1.5GB of RAM on the Raspberry Pi 5. The project includes downloadable STL files for the enclosure alongside the open-source software, allowing users to recreate the hardware.