Voice Assistant for Blind

Abstract

Navigating safely through everyday environments can be extremely challenging for people with visual impairments. To address this, we developed a smart Blind Voice Assistant that uses the YOLOv8 object detection model to identify surrounding objects in real time. The system can detect both common and potentially dangerous items, estimate their approximate distance in steps, and immediately inform the user through voice feedback. This approach allows users to become more aware of their surroundings and make safer decisions as they move around. By training the model with both standard COCO data and a custom dataset including items like knives, pens, and bags, we ensure the assistant is tuned to recognize important objects in daily life. Overall, the assistant aims to empower visually impaired individuals by offering real-time environmental awareness and guided navigation

Introduction

1. Introduction

Visually impaired individuals face significant challenges in navigating environments safely and independently. Traditional tools like white canes and guide dogs are helpful but limited. This project introduces a Blind Voice Assistant that leverages YOLOv8, a state-of-the-art object detection model, to identify and announce objects in real time. It aims to be an affordable, real-time, and offline system for helping blind users with environmental awareness and mobility.

2. Literature Survey

The existing solutions for visually impaired users include:

• Google TalkBack: Screen-reader-based aid, but limited to on-screen interaction.

• DeepNAVI (2023): Context-aware navigation, works offline but lacks full real-time spatial mapping.

• Shen et al. (2022): Used YOLOv3 with haptic feedback but lacks audio alerts.

• Litoriya et al. (2023): Combined YOLO + SSD for better detection but lacks custom object focus.

• Other systems either lack motion sensing, spatial orientation, or require pre-mapped environments.

Gap Identified: No single system provides real-time object detection, spatial localization, voice feedback, and indoor navigation all in one, especially in offline, low-cost setups.

3. Proposed System

A. Overview

A voice assistant for blind users that:

• Uses YOLOv8 for real-time object detection.

• Offers audio feedback via pyttsx3.

• Recognizes both common (COCO dataset) and custom objects (e.g., pen, knife).

• Operates offline, with only a webcam and laptop.

B. System Framework

• OpenCV: Captures real-time frames from webcam.

• YOLOv8: Detects and classifies objects instantly.

• Custom Dataset: Adds domain-specific object recognition.

• pyttsx3: Converts detected data into offline voice output.

• Spatial Mapping: Divides camera view into Left, Center, Right zones and estimates distance in steps.

C. Workflow

1. Initialize system and load models.

2. Capture webcam feed using OpenCV.

3. Detect objects with YOLOv8.

4. Classify danger levels, map spatial zones.

5. Generate context-aware audio messages, e.g., “Knife 3 steps ahead.”

6. Speak feedback via pyttsx3.

7. Repeat until user stops or exits.

D. Key Technologies

• YOLOv8: Fast (1 ms/frame), accurate, anchor-free detection.

• OpenCV: Real-time frame processing.

• pyttsx3: Offline TTS engine.

• Custom Dataset: Tailored to critical, real-world objects.

• Python: Main language for integration.

4. Output

The system demonstrates:

• Object detection in real-time (e.g., books, pens, bottles).

• Voice alerts for proximity and danger.

• Visual previews (optional) via GUI.

• Navigation assistance with spatial warnings (e.g., "Obstacle on your left").

Conclusion

The \"Voice Assistant for Blind\" project seeks to create a more inclusive environment for the visually impaired by integrating computer vision, deep learning, and text-to-speech technologies. Utilizing YOLOv8 for real-time object detection and both default and custom datasets, the system provides customized auditory feedback to assist users with navigation in unfamiliar or dynamic environments. Its capability ensures consistent performance independent of internet connectivity, and a webcam-based visual sensor, optimized processing, and voice or button interface ensure it\'s both functional and easy to use. From detecting common objects in daily life to alerting one to possible threats, the assistant serves as an electronic companion that facilitates mobility, safety, and independence—testifying to the significant role that artificial intelligence can have in assistive technology.

References

[1] Bineeth Kuriakose, Raju Shrestha, Frode Eika Sandnes, “DeepNAVI: A deep learning-based smartphone navigation assistant for people with visual impairments,” Expert Systems with Applications, vol. 212, 2023, Article ID 118720. [2] Shivam Verma, Rohit Omanwar, Sreejith V, Meera G S, “A Smartphone Based Indoor Navigation System,” Proceedings of the IEEE International Conference on Mechatronics (ICM). [3] D. Ravi Kumar, Hiren Kumar Thakkar, Suresh Merugu, Vinit Kumar Gunjan, Suneet K. Gupta, “Object Detection System for Visually Impaired Persons Using Smartphone,” Lecture Notes in Electrical Engineering, vol. 761, Springer, 2022, pp. 1561–1572 [4] Ping-Jung Duh, Yu-Cheng Sung, Liang-Yu Fan Chiang, Yung-Ju Chang, Kuan-Wen Chen, “V-Eye: A Vision-based Navigation System for the Visually Impaired,” IEEE Transactions on Multimedia, 2020. [5] Fanghao Song, Zhongen Li, Brian Clark, Dustin Grooms, Chang Liu, “Camera-Based Indoor Navigation in Known Environments with ORB for People with Visual Impairment,” 2020 IEEE Global Humanitarian Technology Conference (GHTC), 2020. [6] Akilesh Salunkhe, Manthan Raut, Shayantan Santra, Sumedha Bhagwat, “Android-based object recognition application for visually impaired,” ITM Web of Conferences, vol. 40, 03001, 2021. [7] Junjie Shen, Yiwen Chen, Hideyuki Sawada, “A Wearable Assistive Device for Blind Pedestrians Using Real-Time Object Detection and Tactile Presentation,” Sensors, vol. 22, no. 12, 2022. [8] Rana Sabah Naser, Meng Chun Lam, Faizan Qamar, B. B. Zaidan, “Smartphone-Based Indoor Localization Systems: A Systematic Literature Review,” Electronics, vol. 12, no. 8, 2023.

Copyright

Copyright © 2025 Sakshi Natuskar, Anisha Sable, Yashkumar Sharma, Simran Shende, Prof. Dr. P.Y. Itankar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.