Interference Noise In Speech Signals Essay

957 Words2 Pages

Interference noise in speech signals is a problem which occurs frequently in speech processing. Interference noise reduces the clearness of speech signal and modulates it. Interference noise may be produced from any type of signal that interferes with the speech signals example acoustical sources for instance ventilation equipment, echoes, crowds. In analog and digital communications, signal-to-noise ratio, often written S/N or SNR, is a measure of signal strength relative to background noise. The ratio is usually measured in decibels (dB). Ideally, the signal-to-noise ratio is greater that 0dB. This means that the speech is louder than the noise. It is important to note that the amount of speech which may be understood, depends on the …show more content…

When the noise signal is louder than the speech signal, low frequency noise is a much more effective modulation. In addition, the noise is able to encapsulate vowels and consonants at high pressure levels. Noise that affects the speech signals are characterized by one of the following: 1. White Noise: A sound or signal in which all audible frequencies have equal intensity. For each frequency, the phase of the noise is not completely certain. The phase may be shifted up to 360o, and its value is not related to the phase for a given frequency value. The white noise strong encapsulating property is related to its broad-band spectrum. White noise has a zero mean, constant variance, and is uncorrelated in time. White noise has a power spectrum which is uniformly spread across all allowable frequencies. In Matlab, w = randn(N) generates a sequence of length N of n(0, 1) ‘Gaussian’ white noise (i.e. with a normal distribution of mean 0 and std 1). Reference: Hartmann Atm S 552 notes, Chapter 6.1-2. 2. Colored …show more content…

All analog-to-digital converters are known to introduce quantization errors to the incoming signal. When an Analog-Digital Converter (ADC) converts a continuous signal into a discrete digital representation, there is a range of input values that produces the same output. That range is called quantum (Q) and is equivalent to the Least Significant Bit (LSB). Quantization error is the difference between input and output. Therefore, the quantization error can be between ±Q/2. www.onmyphd.com/?p=quantization.noise.snr 3.3.2 Truncation noise If we multiply two n-bit numbers together, one requires 2n bits to store the answer. Therefore, in all fixed-point processors the product register and the accumulator are double the length of all the other registers. When we work with a 32-bit product, after the product is added to the accumulator, the32 bits are maintained during the filter subroutine. The 32-bit result is then stored in a16-bit wide memory. If we use two instructions, the processing time and the amount of memory needed will be doubled. The time required to recover the value of use in future operations will also be increased. Therefore, it is usual to store only the most significant 16bits and truncate the result of the

Open Document