The problem Echo of Words solves

Silent Voice is an AI-powered communication tool for people living with aphasia — a condition (often caused by stroke or brain injury) where a person can think and understand language normally, but struggles to speak or write.

Instead of requiring typing or speech, Silent Voice lets a user build a message by tapping simple icons (emotions, body sensations, needs). An AI model then converts that icon sequence into a natural, complete, warm sentence — which is displayed on screen and read aloud — so the person can be understood by family members, caregivers, or medical staff without needing to produce the words themselves.

Challenges we ran into

• When converting from the localhost URL to the public URL, the generated AI output was different from what we expected. We checked the project from beginning to end and discovered that we had not pulled the latest commit from Git, which caused the issue. Pulling the latest changes fixed the output.
• We also found that generating the output initially took a long time. To improve this, we modified the AI prompt and added restrictions to prevent unnecessary processing, making the generation more efficient.