The use of visual information from speaker's mouth region have shown to improve the performance of automatic speech recognition (ASR) systems. This is particularly important in presence of noise which even in moderate form severely degrades the speech recognition performance of systems using only audio information. Various sets of features extracted from speaker's mouth region have been used to improve upon the performance of an ASR system in such challenging conditions and have met many successes. To the best of authors knowledge, the effect of using these techniques on recognition performance on the basis of phonemes have not been investigated yet. This paper presents a comparison of phoneme recognition performance using visual features extracted from mouth region‐of‐interest using discrete cosine transform (DCT) and discrete wavelet transform (DWT). New DCT and DWT features have also been extracted and compared with the previously used one. These features were used along with audio features based on Me...

/pdf/a-comparison-of-visual-features-for-audiovisual-automatic-qyroaykh0n.pdf

A comparison of visual features for audiovisual automatic speech recognition

Learning Control of Dynamical Systems Based on Markov Decision Processes:Research Frontiers and Outlooks

Acoustic noise has a pronounced negative impact on human vocal communication and user comfort. Interfering noise that is transmitted between communicating users may not only disturb telephone conve ...

/pdf/applied-methods-to-combat-noise-in-human-communication-sqy40csoq9.pdf

Applied Methods to Combat Noise in Human Communication

Orthogonal frequency division multiplexing (OFDM) is characterised by a large peak-to-average-power ratio and an orthogonal signal design. As a consequence, OFDM is particularly vulnerable with respect to nonlinear distortions caused, for example by a nonlinear high power amplifier (HPA). In the presence of nonlinear high power amplification, OFDM suffers from crosstalk between all subcarriers. Furthermore, the power density spectrum is significantly widened.



In this article, we simultaneously use a memoryless predistorter (PD) at the transmitter side and a nonlinear detector at the receiver side. The PD reduces the out-of-band power, whereas the nonlinear detector improves the bit error rate (BER). Since maximum-likelihood detection is prohibitive, a novel reduced-state symbol detector (RSSD) derived from the maximum-likelihood detector is proposed. For a QPSK/OFDM system in the presence of a memoryless solid-state power amplifier, it is shown that the PD is not only useful for spectral shaping, but also in order to reduce the computational complexity of the nonlinear detector and to provide more robustness, concerning an incomplete knowledge of the characteristics of the nonlinearity at the receiver. The overall raw bit error performance is shown to be close to that of a linear QPSK/OFDM system. Copyright © 2007 John Wiley & Sons, Ltd.

Predistortion and nonlinear detection for OFDM signals in the presence of nonlinear high power amplification

Adjacent downlink satellite interference usually occurs because of the presence of side lobes in addition to the desired main lobe in the radiation pattern of the satellite ground station antenna. If the angular separation between two adjacent satellites is not sufficiently large, it is quiet possible that the power radiated through the antenna's radiation pattern, whose main lobe is directed towards the intended satellite, interferes with the received signal from the adjacent satellite. This problem becomes more difficult if both satellites transmit at the same frequency. The SARSAT (Search and Rescue Satellite Aided Tracking) system is designed to provide distress alert and location data in order to assist in search and rescue operation. SARSAT uplink detects and locates distress beacons (406MHz) activated at distress location. The system calculates the location of distress event using Doppler processing techniques. Processed data is repeatedly retransmitted through SARSAT downlink to Local User Terminals (LUT) when satellites are in view and also stored in on-board memory. The downlink of all SARSAT LEO (Low Earth Orbit) satellites uses the same 1544.5 MHz frequency. In case those satellites are within the main lobe of the local user terminal antenna, transmissions from adjacent satellites act as interference to each other. This can result in missed bursts or no solutions received at LUT, consequently no data provided about the distress location∗. Modulation index application for adjacent satellite downlink interference identification, for periods where the interference duration may be significant, is presented in this paper. Measurement results recorded at SARSAT terminals are provided.

Modulation index application for satellite adjacent downlink interference identification

In this paper, a robust speech recognition system in the presence of additive background noise is studied. We adopt speech enhancement as the front-end processing module to improve the signal-to-noise ratio (SNR) of the input noisy speech for recognition in the latter stage. Three different speech enhancement algorithms were tested on six types of additive noises. Experimental results show that the cascading of the speech enhancer and a hidden Markov model (HMM) based speech recognizer can significantly improve recognition accuracy in noisy environments.

Robust recognition of noisy speech using speech enhancement

From the viewpoint of decision theory, AQM (active queue management) can be considered as an optimal decision problem. In this paper, a new AQM scheme, Reinforcement Learning Gradient-Descent (RLGD), is described based on the optimal decision theory of reinforcement learning. Aiming to maximize the throughput and stabilize the queue length, RLGD adjusts the update step adaptively, without the demand of knowing the rate adjustment scheme of the source sender. Simulation demonstrates that RLGD can lead to the convergence of the queue length to the desired value quickly and maintain the oscillation small. The results also show that the RLGD scheme is very robust to disturbance under various network conditions and outperforms the traditional REM and PI

/pdf/a-robust-active-queue-management-algorithm-based-on-4lptdc961l.pdf

A Robust Active Queue Management Algorithm Based on Reinforcement Learning

Modulation index is an important parameter of CPFSK (continuous phase frequency shift keying) signals, which affects the system performance significantly. In noncooperative communication systems, this index is unknown to the receiver and must be estimated. This paper proposed a fast modulation index estimation algorithm for CPFSK signals based on short time Fourier transform (STFT). Simulations show the estimation results fit the actual value well. Furthermore, this paper evaluated the performance of timing synchronization and BER using the estimated modulation index, which is near to the performance using the real index.

Modulation index estimation for CPFSK signals and its application to timing synchronization

This paper presents a new method of nonlinear distortion compensation for OFDM systems. The proposed scheme compensates for the performance degradation due to the nonlinear distortion by using the combination of predistortion and reconstruction. The computer simulation results show that the proposed scheme can compensate for degradation due to the nonlinear distortion significantly. Furthermore, since the proposed scheme utilizes the iterative method, its implementation is relatively simple.

Compensation of nonlinear distortion using a new method combining predistortion with reconstruction in OFDM systems

Continuous phase frequency shift keying (CPFSK) is a power and bandwidth efficient modulation. This paper proposes a novel method for carrier recovery in CPFSK demodulation. Its complexity is very low, and it is also easy to implement, especially for software implementation. The proposed method has an excellent bit error rate performance.

http://ieeexplore.ieee.org/iel5/7138/19221/00890915.pdf

Ma Zhengxin

Papers

Robust recognition of noisy speech using speech enhancement

A Robust Active Queue Management Algorithm Based on Reinforcement Learning

Modulation index estimation for CPFSK signals and its application to timing synchronization

Compensation of nonlinear distortion using a new method combining predistortion with reconstruction in OFDM systems

A novel carrier recovery method for CPFSK demodulation