fasuizu-br

Speech AI MCP server with pronunciation assessment (phoneme-level scoring, 17MB model, <300ms), text-to-speech, and speech-to-text. 8 tools for AI agents building language learning, accessibility, and voice applications.