ASR Mobile Application with 6G Beamforming Simulation

Abstract

Most speech analysis tools today either work only on desktops or need a constant internet connection, which makes them hard to use for students and researchers on the go. On the wireless side, understanding how 6G beamforming works is tough without expensive hardware. To tackle both these issues, we built a project that brings together three things in one package. First, a Flutter-based mobile app that lets users record audio, view waveforms, run offline speech-to-text using Whisper, and generate spectrograms and pitch graphs. Second, a Python-based 6G beamforming simulation that compares square and star antenna layouts using CNN-based beam prediction along with classical methods. Third, a simple ESP32 and servo hardware demo that physically shows how beam scanning and angle selection work. Testing showed that all three modules perform reliably and complement each other well for academic use.

Introduction

The text describes a multidisciplinary project that combines mobile speech analysis, 6G beamforming simulation, and a simple hardware demonstration to help students learn advanced concepts in signal processing and wireless communication.

The motivation behind the project is that modern smartphones can now perform complex tasks like speech recognition and signal analysis, while 6G research involves advanced beamforming techniques that are difficult for students to experiment with due to expensive hardware. To solve this, the project integrates three parts: a Flutter-based offline speech analysis app, a Python-based beamforming simulation, and an ESP32-based physical beam scanning model.

The speech module allows users to record audio, visualize waveforms, and perform offline speech-to-text using Whisper, along with generating spectrograms, pitch, and intensity graphs. The beamforming module simulates a 64-element antenna array using machine learning (CNN) and compares it with traditional methods like MRT and ZF. The hardware module uses an ESP32, servo motor, and ultrasonic sensor to physically demonstrate directional scanning similar to real beamforming systems.

The system works by allowing users to analyze speech data on mobile, simulate beam selection in software, and observe real-world scanning through hardware. It is fully offline, cost-effective, and suitable for educational use.

Testing results show that the speech app works accurately, the simulation produces meaningful comparisons between AI-based and classical beamforming methods, and the hardware successfully demonstrates directional scanning.

Conclusion

In this project, we put together a mobile speech analysis app, a 6G beamforming simulation, and a physical beam scanning demo into one coherent package. Each part works well on its own, and together they form a solid academic platform. The speech app handles recording and analysis offline, the simulation compares antenna geometries with real depth, and the hardware makes the whole thing tangible. Future work could include adding a camera feed, richer acoustic features, and tighter app-to-hardware communication.

References

[1] A. Radford, J. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust Speech Recognition via Large-Scale Weak Supervision,” OpenAI Technical Report, 2022. [2] P. Wang and Y. Li, “Beamforming Techniques for 5G and Beyond Wireless Systems,” IEEE Communications Surveys and Tutorials, vol. 25, no. 2, pp. 834–856, 2023. [3] M. Bhuyan and S. Das, “Adaptive Hybrid Beamforming Using Reinforcement Learning,” IEEE Access, vol. 12, pp. 22145–22158, 2024. [4] S. N. Awan, “Validity of Acoustic Measures Obtained Using Smartphones for Voice Analysis,” Journal of Voice, vol. 38, no. 3, pp. 612–620, 2024.

Copyright

Copyright © 2026 Dr. G. Balachandran, Aadhil J, Kiran P. 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.