Aseto.ai Speech-to-Text Model Development: Advancing Voice AI for Cypriot and Greek Languages
Aseto.ai Speech-to-Text Model Development: Advancing Voice AI for Cypriot and Greek Languages

Aseto.ai has developed an advanced Speech-to-Text (STT) model purpose-built for the Cypriot dialect and the Greek language, achieving a substantial reduction in transcription errors compared to leading open models. This work addresses a long-standing limitation in Voice AI technology for regional languages, enabling practical deployment of Cypriot and Greek voice agents across business and consumer applications. Early benchmarking shows over 30% relative improvement in Word Error Rate (WER) for Cypriot speech and over 40% improvement for standard Greek while maintaining strong English performance. These results demonstrate that regional optimization can deliver tangible accuracy gains without compromising multilingual capability.