Reproduce “LUMO (Lightweight Unified Multilingual Orchestrator): A Privacy Preserving Offline Voice Assistant” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Md. Mehedi Hasan Naeem, Mst. Kamrunnahar Ruma, Nafiza Anjum, Shakila Sultana, Md. Sujan Ali
LUMO integrates automatic speech recognition, local language-model response generation, and text-to-speech on a Raspberry Pi 5, using voice activity detection, VOSK, a quantized local language model, and Piper in an offline English/Bangla pipeline. On 1,000 short command-style utterances, it reports 6.8% WER at 50 dB and 3.2-second response latency. These measurements do not establish broad conversational competence or comparative superiority.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.