lezu
cycterna2222288888-ai/lezu
Open dataset + benchmark for Armenian-Russian code-switched speech recognition
speech-recognitioncode-switchingarmenian-languagerussian-languagedatasetbenchmarkmachine-learningforced-alignmentmultilingual-asr
Summary
Lezu is an early-stage research project aiming to create the first open dataset and benchmark for Armenian-Russian code-switched speech recognition. The repository currently contains only a detailed README outlining the problem, related work, and a multi-stage research plan. No data or code has been produced yet, as the project is in the scoping and planning phase.