A CNN-based Vocal Recognition Algorithm for IARC Mission 8

摘要

In this paper, a vocal command recognition problem is investigated to meet the rule of the International Aerial Robot Competition (IARC) Mission 8, which requires the participant to interact with UAVs through non-electrical methods. First, the audio signal description is given, and the preprocessing of the data is provided. Then, a CNN-based vocal recognition algorithm with model of residual connection is proposed in which feature extraction with preprocessing method is applied instead of taking the raw data as the input. Moreover, the proposed algorithm is implemented on an android platform. Then, simulation experiments are conducted to show the effectiveness of the proposed algorithm. Besides, comparisons and analysis are given for different methods of data processing.