S Jakati Dr.Shridhar.K
Institute of Techonology
Speech enhancement aims to improve speech quality by using various
algorithms. The objective of enhancement is
improvement in intelligibility or overall perceptual quality of degraded speech signal using audio signal
processing techniques. Speech
enhancement is necessary for many applications in which clean speech signal is
important for further processing. The speech enhancement techniques mainly
focus on removal of noise from speech signal. The various types of noise and
techniques for removal of those noises are presented in speech signal. Most
widely used speech enhancement technique namely, FIR,IIR and Multirate method is reviewed in this paper with its
state-of-art for better noise cancellation.
Key words: Speech enhancement, FIR,IIR and Multirate
Speech enhancement is an area of
speech processing where the goal is to improve the intelligibility or
pleasantness of a speech signal. The most common approach in speech enhancement
is noise removal, where we, by estimation of noise characteristics, can cancel
noise components and to get the clean speech signal. The basic problem with
this approach is that if we remove those parts of the signal that resemble
noise. In other words, speech enhancement algorithms, Current speech processing
algorithms can roughly be divided into three domains, spectral subtraction,
sub-space analysis and filtering algorithms:
• Spectral subtraction algorithms operate in the
spectral domain by removing, from each spectral band, that amount of energy
which corresponds to the noise contribution. While spectral subtraction is
effective in estimating the spectral magnitude of the speech signal, the phase
of the original signal is not retained, which produces a clearly audible
distortion known as “ringing”.
• Sub-space analysis operates in the autocorrelation
domain, where the speech and noise components can be assumed to be orthogonal,
whereby their contributions can be readily separated. Unfortunately, finding
the orthogonal components is computationally expensive. Moreover, the orthogonality
assumption is difficult to motivate.
• Finally, filtering algorithms are time-domain
methods that attempt to either remove the noise component or estimate the noise
and speech components by a filtering approach (IIR,FIR filtering and multirate
2. IIR FILTER
IIR filters are
digital filters with infinite impulse response. They have the feedback (a
recursive part of a filter) and are known as recursive digital filters because
fraction of the output signal is fed
back to the input.IIR filters have much better frequency response than FIR
filters of the same order. Unlike FIR filters, their phase characteristic is
not linear which can cause a problem to the systems which need phase linearity.
For this reason, it is not preferable to use IIR filters in digital signal
processing when the phase is essence.
There is one problem
known as a potential instability that is typical of IIR filters only. FIR
filters do not have such a problem as they do not have the feedback. For this
reason, it is always necessary to check after the design process whether the
resulting IIR filter is stable or not. The input X(n) and the output
Y(n)of a causal IIR filter satisfy the
linear constant-coefficients difference equation of the form.
Advantages and Disadvantages
The main advantage digital IIR filters have over FIR
filters is their efficiency in implementation, in order to meet a specification
in terms of passband, stopband, ripple, and/or roll-off. Such a set of
specifications can be accomplished with a lower order IIR filter than would be
required for an FIR filter meeting the same requirements.
FIR filters can be easily made to be linear phase, a property that is not easily met
using IIR filters and then only as an approximation.
Digital IIR filters are the potential for limit cycle behavior when idle, due to
the feedback system in conjunction with quantization.
response (FIR) filter is a filter whose impulse response (or response to any finite
length input) is of finite duration, because it settles to zero in
finite time. An FIR filter
is usually implemented by using a series of delays, multipliers, and adders to
create the filter’s output.
Figure 2 shows the basic block diagram for an
FIR filter of length N. The delays result in operating on prior input samples.
The h0 values
are the coefficients used for multiplication, so that the output at time n is
the summation of all the delayed samples multiplied by the appropriate
coefficients. A causal FIR filter
has the following difference equation
is the order. The result y(n) is the discrete convolution of x(n) with the
(finite) impulse response:
Advantages and Disadvantages
Ø There is no feedback loop in the structure of
an FIR filter. Due to not having a feedback loop, an FIR filter is inherently
stable. Meanwhile, for an IIR filter, we need to check the stability.
Ø FIR filter can provide a linear-phase response.
As a matter of fact, a linear-phase response is the main advantage of an
FIR filter over an IIR design otherwise, for the same filtering specifications,
an IIR filter will lead to a lower order
Ø The main disadvantage of FIR filters is that
considerably more computation power in a general purpose processor is required
compared to an IIR filter with similar sharpness or selectivity especially
when low frequency (relative to the sample rate) cutoffs are needed.
Ø Digital signal processors provide specialized
hardware features to make FIR filters approximately as efficient as IIR for
channel version, popularly called the Quadrature mirror Filter Bank. The input
signal X(n) is first filtered by two
filters H0(z) and H1(z),typically lowpass and high pass
filter. Each signal Xk(n) sub band signal with band limited to width
of ?. The subband signal are decimated by factor of 2 to produce Vk(n).Each
decimated signal Vk(n) is then coded in special properties of the
sub band are exploited. At the receiver end the received signal are decoded to
produce the signal V0(n) and V1(n) which are passed
through two-fold expanders. The output signal Y0(n) and Y1(n)
are then passed rough the filters F0(z) and F1(z) to
produce the output signal X^(n)
Ø H0(z) and H1(z)-Analysis
Ø V0(n) and V1(n)-
Ø F0(z) and F1(z) –
Ø Y0(n) and Y1(n)-Expanded
Ø X^(n)-Reconstructed signal
Compared to IIR filters, FIR
filters and Multirate Systems.