System / 07

Spoken user guidance

Turns measurable image problems into short instructions that help improve the next frame.

PIPELINE STATEfeedback loop

Purpose

Close the loop when the user cannot visually verify camera position or image quality.

Role in the complete pipeline

Guidance interprets blur, glare, distance, page visibility, and stability signals before recognition is attempted again.

INPUT

Quality measurements, detection state, OCR validation, and prompt history.

METHOD
  1. Prioritize the most actionable issue
  2. Map it to a short spoken instruction
  3. Apply cooldowns to prevent repetition
  4. Confirm when the page is ready
OUTPUT

Prompts such as “Move closer,” “Hold steady,” “Reduce glare,” or “Ready to read.”

Common failure cases

  • Prompts repeat too often
  • Several problems compete for attention
  • A wrong diagnosis makes capture worse
  • Instructions arrive too late

How VisoraAI responds

  • Speak one issue at a time
  • Use prompt cooldowns
  • Stop prompting once the frame is acceptable

Current limitations

  • Thresholds need user testing
  • Not every failure can be translated into a simple correction

Planned improvements

  • Evaluate prompt clarity with users
  • Tune priorities by failure type
  • Add nonverbal alignment cues where appropriate