Millisecond-scale detection of mouse ultrasonic vocalizations enables closed-loop experiments
Mouse ultrasonic vocalizations (USVs) provide a rapidly evolving readout of social interaction but are typically analysed only after acquisition. Here we introduce DeepFisFis, a waveform-based neural network that detects USVs while they are being produced. DeepFisFis classifies consecutive 5 ms audio segments directly from the waveform with high accuracy, processing each segment in approximately…
Mouse ultrasonic vocalizations (USVs) are used to gauge social interactions, but they are usually analyzed after recording. Researchers have now developed DeepFisFis, a neural network that detects USVs while they are being made. DeepFisFis examines 5 ms audio segments directly from the waveform, classifying them with high accuracy.
Each segment is processed in about 2.5 ms, which is quicker than the incoming audio stream. This real-time detection allows ongoing vocalizations to influence experimental interventions instantly.
In a practical application, the detections from DeepFisFis triggered an external stimulus, showing the system's ability to control ongoing vocal behavior online. DeepFisFis also enables event-triggered data acquisition. By gating storage around detected calls, the system managed to keep over 99% of vocalization time while only storing about 22% of the continuous recording.
This innovation turns USVs from a post-experiment behavioral readout into a real-time experimental signal. With DeepFisFis, researchers can selectively acquire vocal communications and explore the underlying neural circuits through causal interrogation.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
