Acoustic Voice-Call Data Relay

During severe socio-political crises, such as the July Movement, authorities frequently mandate complete internet blackouts and block SMS-gateways to suppress information flow. This absolute communication isolation renders traditional messaging platforms—like WhatsApp, Messenger, and standard SMS—completely useless. Citizens, activists, and emergency responders find themselves dangerously disconnected during critical moments when coordination is a matter of life-and-death. Even when mobile networks remain physically intact, restrictive state-sponsored blackouts strip away all modern digital data channels. However, standard 2G-voice calls typically remain operational, as completely disabling voice infrastructure is often unfeasible or causes unacceptable collateral damage. The core problem is that this remaining voice channel is strictly designed for human speech, not digital data transfer. We lack a reliable, infrastructure-independent method to transmit critical digital information (like GPS coordinates or encrypted messages) across these surviving low-bandwidth voice networks during total internet shutdowns.

Our solution, the Acoustic Voice-Call Data Relay, completely bypasses internet shutdowns and SMS blocks by transmitting digital data as sound over standard 2G voice calls. When traditional data infrastructure fails, our system leverages the one communication channel that consistently remains online: basic cellular voice networks. Instead of relying on Wi-Fi or mobile data, our software encodes critical information—such as text messages, GPS coordinates, or even small image files—into audio signals using Frequency Shift Keying (FSK). The encoded audio is transmitted across the cellular network just like human speech. On the receiving end, our custom Digital Signal Processing (DSP) pipeline captures the audio and perfectly decodes it back into the original digital data. Building this required overcoming severe technical challenges, specifically the aggressive audio compression algorithms (AMR vocoders) used by telecom networks which are optimized exclusively for human voices. We engineered a highly robust DSP-pipeline that utilizes custom packet framing, synchronization patterns, and Forward Error Correction (FEC). This ensures our data survives heavy network distortion, background noise, and packet-loss. Our testing proves the system successfully decodes data even in incredibly hostile cellular environments down to -14dB Signal-to-Noise Ratio. The current implementation features an interactive, easy-to-use interface that supports automatic image downsampling, byte-for-byte file reconstruction, and live channel corruption simulations. Looking ahead, our final product vision integrates this DSP pipeline directly into open-source dialers (like Fossify Phone) via Chaquopy or as messenger plugins. Furthermore, we are designing a Peer-to-Peer (P2P) SOS feature that can autonomously bounce emergency beacons across interconnected voice calls, creating a decentralized mesh-network of survival during absolute blackouts. By transforming a primitive voice channel into a resilient data modem, we provide a vital, unblockable lifeline during the most critical times.

numpy, scipy, reedsolo, sounddevice, streamlit, matplotlib, pytest, rnnoise

Gemini, Claude