II
Empirical mode decomposition and the deep belief network in the
application of speech recognition research
ABSTRACT
With the rapid development of intelligent applications and networking technology, the
speech recognition as a convenient and effective human-computer interaction whose importance
becomes increasingly prominent. However, due to the complicated environment of the speech
production and huge difference in human language accent, the speech recognition accuracy faces
greater difficulties. So it is difficult to achieve the expected experience results about the
corresponding speech products. To make the speech recognition technology breakthrough, it
must make an improvement and innovation in speech recognition on the basis of theory.
As the speech signals are typically nonlinear and non-stationary, the traditional speech
processing methods such as Fourier transform, Wavelet transform are assumed that the signal is
smooth short. Therefore, the conclusions are bound to disrupt the analysis and extraction of the
useful information. As the background, this paper describes the basics of speech recognition as
the premise, then it focuses on the speech endpoint detection algorithm under different SNR and
the isolated word speech recognition system based on deep belief network.
The main contents and innovations in this paper are as follows:
(1) In-depth study of the limitations of the traditional speech processing methods, this paper
focuses on the empirical mode decomposition as a new signal processing method, and it makes a
algorithm verification about the character which the empirical mode decomposition is suitable
for nonlinear and non-stationary signal.
(2) As the traditional speech endpoint detection algorithms exist a shortcoming which the
speech signal detection accuracy is not high for the noisy speech, this paper proposes a new
speech endpoint detection algorithm based on empirical mode decomposition and composite
energy. Firstly, this algorithm uses empirical mode decomposition to make the speech signal
decomposed into a set of intrinsic mode functions and a remaining amount, and it filters out the
low-level intrinsic mode functions which contain a lot of noise, the remaining intrinsic mode
functions are reconstructed into the speech signal. Then it uses the character of Teager energy
about compression noise amplitude to calculate the each frame’s Teager energy of the
reconstructed speech signal, so it can obtain a composite energy per frame of the signal by
weighting the Teager energy to the short-term energy. Finally, it can get the starting and ending
points of the signal by comparing the composite energy and adaptive threshold for each frame.
The simulation results show the effectiveness of the algorithm, compared to the traditional
万方数据
