Frequency-Domain Voice Activity Detection

This model detects voice activity using a frequency-domain audio signal.

Voice Activity Detection is often used as an indication whether further processing or analysis of a signal is required. Many processing and analysis techniques require a frequency-domain representation of the signal. For example, the voice activity detection algorithm operates in the frequency domain. To save computation, you can convert the audio signal to the frequency domain once, and then feed the frequency-domain signal to downstream analysis and processing.

This model additionally buffers the signal so that the VAD operates on half-overlapped frames. Overlapping the input frames to the VAD increases the accuracy and resolution in time of the probability of speech.