Word Error Rates (WER) for isolated digit recognition task reported in the literature for different recognition methods.

Note: DEM (Dynamic Expectation Maximization) is the recognition system used in this paper; LSTM (Long Short-Term Memory) network was introduced in [111], LSM (Liquid State Machine) with 1232 neurons was reported in [51] and was improved (LSM 2) in [18]. The results for the state-of-the-art speech recognition system using HMM (Hidden Markov Model) were reported in [18]. OT (Occurrence Time) features were used in [103].