Back to Projects


Hearsona
Thesis Project • 2025
PythonReactMistral 7B InstructAudioLDM2Websockets
Overview
Engineered a Human-in-the-Loop (HITL) generative AI pipeline that empowers users to co-create personalized auditory cues. The system integrates Mistral 7B for intent reasoning and AudioLDM2 for synthesis, allowing users to iteratively refine outputs through natural language feedback and fine-grained controls. This interactive workflow ensures the generated cues are not only high-fidelity but semantically aligned with the user’s specific cognitive associations.
Gallery
Interface
HITL conversation
Hearsona Pipeline
Audio Generation Demo