Radios and cellphones use frequency modulation (FM) of an oscillating carrier signal to reliably transmit multiplexed data while rejecting noise. Here, we establish a biochemical analogue of this paradigm using genetically encoded protein oscillators (GEOs) as carrier signals in circuits that enable continuous, real-time FM streaming of single-cell data. GEOs are constructed from evolutionarily diverse MinDE-family ATPase and activator modules that generate fast synthetic protein oscillations when co-expressed in human cells. These oscillations serve as a single-cell carrier signal, with frequency and amplitude controlled by GEO component levels and activity. We systematically characterize 169 ATPase/activator GEO pairs and engineer composite GEOs with multiple competing activators to develop a comprehensive platform for waveform programming. Using these principles, we design circuits that modulate GEO frequency in response to cellular activity and decode their responses using a calibrated machine-learning model to demonstrate sensitive, real-time FM streaming of transcription and proteasomal degradation dynamics in single cells. GEOs establish a dynamically controllable biochemical carrier signal, unlocking noise-resistant FM data-encoding paradigms that open new avenues for dynamic single-cell analysis.