A New Method for Low SNR Estimation of Noisy Speech Signals Using Fourth-Order Moments
スポンサーリンク
概要
- 論文の詳細を見る
In this paper the SNR estimation is performed frame by frame, during the speech activity. For this purpose, the fourth-order moments of the real and imaginary parts of frequency components are extracted, for both the speech and noise, separately. For each noisy frame, the mentioned fourth-order moments are also estimated. Making use of the proposed formulas, the signal-to-noise ratio is estimated in each frequency index of the noisy frame. These formulas also predict the overall signal-to-noise ratio in each noisy frame. What makes our method outstanding compared to conventional approaches is that this method takes into consideration both the speech and noise identically. It estimates the negative SNR almost as well as the positive SNR.
- （社）電子情報通信学会の論文
- 2010-06-01
著者
-
Shamsi Hossein
Electrical Faculty K.n. Toosi University Of Technology
-
Doost Roghayeh
Electrical Engineering Department University Of Amirkabir
-
Sayadian Abolghasem
Electrical Faculty Amirkabir University Of Technology
-
Doost Roghayeh
Electrical Faculty Amirkabir University Of Technology
-
Shamsi Hossein
Electrical Faculty K. N. Toosi University Of Technology
関連論文
- A new perceptually weighted distance measure for vector quantization of the STFT amplitudes in the speech application
- Analysis of the Clock Jitter Effects in a Time Invariant Model of Continuous Time Delta Sigma Modulators( Analog Circuit Techniques and Related Topics)
- A New Method for Elimination of the Clock Jitter Effects in Continuous Time Delta-Sigma Modulators(Nonlinear Theory and its Applications)
- A New Method for Low SNR Estimation of Noisy Speech Signals Using Fourth-Order Moments
- A High PSRR Bandgap Voltage Reference with Virtually Diode-Connected MOS Transistors
- Jitter-Induced Noise Spectrum at the Output of Continuous-Time ΔΣ Modulators with NRZ Feedback Waveform
- Less jitter sensitive NTF design for NRZ multi-bit continuous-time Delta-Sigma modulators