During January 2025, Stephen focused on backend development for the abeiro/HerikaServer repository, delivering a KoboldCPP Text-to-Speech integration. He implemented PHP logic to process TTS requests, enabling configurable endpoints and voice selection through the KoboldCPP API. To optimize performance and reduce reliance on external services, Stephen developed a dedicated PHP module that manages interactions with the TTS service and introduced audio caching, allowing previously generated audio to be reused efficiently. His work combined API integration, backend engineering, and text-to-speech technologies, resulting in a robust feature that streamlines TTS processing and minimizes latency for downstream applications using HerikaServer.