Each sound event takes its own path through the room.
Christian Marclay's "Sound Event" would involve rolling quantities of small balls (golf, Ping-Pong, tennis) down the ramp.
Figure 7 Block-wise accuracy for sound event detection.
SVM-based methods show high performance on sound event recognition.
Based on the recognized context, a context-specific set of sound event classes is selected for the sound event detection stage.
The proposed sound event detection system is composed of two stages: a context recognition stage and a sound event detection stage.
Human-assisted sound event recognition contains three functions: sound source position estimation, human-assisted labelling, and autonomous sound event recognition as shown in Fig. 2.
Figure 1 Training material containing overlapping sound events is used to train both sound event models.
Automatic sound event detection systems are usually designed for specific tasks or specific environments.
A model network is shown in Figure 2. Figure 2 Fully connected sound event model network.
Thus, we propose a sound event detection system utilizing the context information.
