This paper explores the integration of Artificial intelligence (AI), in the form of advanced real time speech to text recognition models integrated to enable a cobot to interpret and act on spoken instructions autonomously. A comparative study across four workstation setups, with voice control was conducted under three key scenarios and variant inputs. Experimental results from RoboDK simulations and physical testbed experiments demonstrate that voice-controlled cobots reduce response times and improve flexibility compared to manual and traditional HRC setups. Emergency stop activation was 40–50% faster, and task adaptation required significantly less intervention. The system supported over 14 voice commands, with recognition rates ranging from 79% to 99%, depending on word complexity and structure. Commands such as “Stop” and “Repeat” achieved near-perfect accuracy while multi-word phrases showed slightly lower performance. Voice-based human confirmations were also reliably detected, contributing to logical task application. Measured inference times across selected commands ranged from 2.31 s to 2.85 s, with Real-Time Factors (RTF) consistently below 1 confirming the system’s suitability for real-time deployment.