Morse Code Audio Recognition using LSTM-CTC Model

Abstract

In media transmission, Morse code is utilized to encode text characters as sequences of two different duration of a signal. Morse code audio is also utilized for SOS of maritime. The popular audio recognition algorithms are Hidden Markov Model, Gaussian Mixture Model, LSTM, and so on. This paper focuses on Morse code audio recognition using LSTM. Long Short-Term Memory (LSTM) is selected because it is the best deep learning algorithm for the sequence-to-sequence model. Morse code audio dataset was created text characters to Morse code audio. Mel-frequency Cepstral Coefficients (MFCC) algorithm is used to extract features of the dataset. LSTM handles to get related information of each Morse code audio by using MFCCs features. Connectionist Temporal Classification (CTC) decides to get the related label by using LSTM information. This paper aims to help the Morse code audio listener by getting the right information about the input data.