Supervised model training for overlapping sound events based on unsupervised source separation

Abstract

Sound event detection is addressed in the presence of overlapping sounds. Unsupervised sound source separation into streams is used as a preprocessing step to minimize the interference of overlapping events. This poses a problem in supervised model training, since there is no knowledge about which separated stream contains the targeted sound source. We propose two iterative approaches based on EM algorithm to select the most likely stream to contain the target sound: one by selecting always the most likely stream and another one by gradually eliminating the most unlikely streams from the training. The approaches were evaluated with a database containing recordings from various contexts, against the baseline system trained without applying stream selection. Both proposed approaches were found to give a reasonable increase of 8 percentage units in the detection accuracy.

Convolutional Recurrent Neural Networks for Polyphonic Sound Event Detection

“…CRNN has significantly higher performance than previous methods [14], [24], [25], [38], and it still shows considerable improvement over other neural network approaches.…”

“…First published systems were scene-dependent, where information about the scene is provided to the system and separate event models are trained for each scene [14], [24], [25]. More recent work [11], [15], as well as the current study, consist of scene-independent systems.…”

“…More recent work [11], [15], as well as the current study, consist of scene-independent systems. Methods [24], [25] are HMM based, using either multiple Viterbi decoding stages or NMF pre-processing to do polyphonic SED. In contrast, the use of NMF in [14] does not build explicit class models, but performs coupled NMF of spectral representation and event activity annotations to build dictionaries.…”

“…This simple architecture-while vastly improving over established approaches such as GMMHMMs [24] and NMF source separation based SED [25], [26]-presents two major shortcomings: (1) it lacks both time and frequency invariance-due to the fixed connections between the input and the hidden units-which would allow to model small variations in the events; (2) temporal context is restricted to short time windows, preventing effective modeling of typically longer events (e.g., rain) and events correlations.…”

Metrics for Polyphonic Sound Event Detection

“…A more complex situation deals with detecting sound events in audio with multiple overlapping sounds, as is usually the case in our everyday environment. In this case, it is possible to perform detection of the most prominent sound event from the number of concurrent sounds at each time [15], or detection of multiple overlapping sound events [16][17][18]. We use the term polyphonic sound event detection for the latter, in contrast to monophonic sound event detection in which the system output is a sequence of non-overlapping sound events.…”