Individual on-line variance adaptation of frequency filtered parameters for robust ASR
نویسندگان
چکیده
In this paper we address the problem of robust speech recognition. We propose a new method based on the individual variance adaptation of frequency filtered parameters to reduce the deleterious effects of additive narrow-band noise. The method can be interpreted as a spectral weighting that assigns increased importance to the most reliable spectral components, typically the spectral peaks. The experiments confirm that the suggested method results in significantly improved recognition rates for additive narrowband noise.
منابع مشابه
Uncertainty decoding on Frequency Filtered parameters for robust ASR
The use of feature enhancement techniques to obtain estimates of the clean parameters is a common approach for robust automatic speech recognition (ASR). However, the decoding algorithm typically ignores how accurate these estimates are. Uncertainty decoding methods incorporate this type of information. In this paper, we develop a formulation of the uncertainty decoding paradigm for Frequency F...
متن کاملRobust tests for testing the parameters of a normal population
This article aims to provide a simple robust method to test the parameters of a normal population by using the new diagnostic tool called the “Forward Search” (FS) method. The most commonly used procedures to test the mean and variance of a normal distribution are Student’s t test and Chi-square test, respectively. These tests suffer from the presence of outliers. We introduce the FS version of...
متن کاملIS-MRAS With On-Line Adaptation Parameters Based on Type-2 Fuzzy LOGIC for Sensorless Control of IM
This paper suggests novel sensorless speed estimation for an induction motor (IM) based on a stator current model reference adaptive system (IS-MRAS) scheme. The IS-MRAS scheme uses the error between the reference and estimated stator current vectors and the rotor speed. Observing rotor flux and the speed estimating using the conventional MRAS technique is confronted with certain problems relat...
متن کاملComparing acoustic features for robust ASR in fixed and cellular network applications
Within the context of automatic speech recognition (ASR) applications for telephony, we investigate the acoustic pre-processing issues that are at stake in going from the xed line to the cellular network. Because the spectral representation used in enhanced full rate GSM is linear prediction, we investigate the relative advantages and drawbacks of conventional mel-frequency cepstral coeecient (...
متن کاملکاهش مصرف سوخت در یک سیستم تغذیه جریان متناوب مستقل با به کارگیری مشخصههای کنترل فرکانس غیرخطی مقاوم
This paper presents nonlinear frequency control characteristic for a dispatchable source to reduce fuel consumption in a standalone AC power system. Power management used in grid is local. The proposed frequency characteristics that are employed for the sources are designed based on grid model. To have robust frequency control characteristics, uncertainty in grid model that consists of load mod...
متن کامل