Alango Acoustic Feedback Cancelation

background image

Page 1 of 5

Acoustic Feedback Cancellation

Alango technology white paper

Revision: 1.2 (23-Oct-2006)


The problem

Acoustic feedback happens in systems where a microphone and a loudspeaker reproducing the
microphone signals are acoustically coupled. A typical situation arising in sound reinforcement
systems is depicted on Figure 1. Other typical cases of acoustic feedback happen in two-way
communication systems and hearing aids.

Amplifier

Figure 1 Audio feedback in a sound reinforcement system

Speaker voice being recorded by a microphone is amplified and reproduced via loudspeaker(s).
Sound produced by the loudspeakers reaches the microphone back either directly or being
reflected from the walls, floor or ceiling. Consequently it enters recording/reproduction cycle
either constantly amplifying itself or eventually dying out. This generally happens on particular
frequencies such that the signal from the loudspeaker arrives to the microphone “just in phase”
to amplify itself. The acoustic feedback becomes a problem when the total system gain on a
specific frequency exceeds the unit value as illustrated on Figure 2.

background image

Acoustic Feedback Cancellation Alango Technologies Ltd. white paper

Page 2 of 5

A(f)

B(f)

Feedback when

G(f)=E(f)

´A(fB(f) >1

Amplifier

A(f)

Equalizer

E(f)

E(f)

Figure 2 The cause of acoustic feedback problem

In such case this frequency in the microphone-loudspeaker-microphone cycle is amplified
more and more, quickly reaching the maximal level of the sound system

Analog solutions

There are several standard “recipes” to reduce the total system gain G by reducing one of its
components A, E or B:

- Reduces the amplification (gain A).

Problem: this may contradict with required loudspeaker power;

- Increase the distance between the microphone and the loudspeaker (gain B).

Problem: not always possible due to physical limitations;

- Use special acoustic treatment of the room (gain B).

Problem: may not be possible due to budget, time or other constraints;

- Reduce the microphone gain (gain A).

Problem: requires closer distance between the user mouth and the microphone. This is
inconvenient, leads to the proximity effect for a directional microphone and causes
puffing and popping sounds during the speech;

- Use equalizer to suppress “problematic” frequencies (gain E on specific frequencies).

Problems: there are may be many of such frequencies; they change if the position of
either speaker or microphone is not fixed;

The advantage of the feedback reduction methods described above is that they don’t need a
digital signal processor. However, all of them provide, at best, only a limited solution that may
be either not satisfactory or not feasible in particular conditions.

background image

Acoustic Feedback Cancellation Alango Technologies Ltd. white paper

Page 3 of 5

Digital solutions

In the digital approach a Digital Signal Processor (DSP) is added into the path as shown on
Figure 3.

EQ

DSP

Amp

G(f)= E(fA(fB(f) ´ D(f)

A(f)

D(f)

E(f)

B(f)

Figure 3 Digital way of feedback suppression

The main advantage of using DSP is in its ability to analyze the signal and perform intelligent
and/or non-linear operations. Of course, DSP may perform equalization as well so that
Equalizer block may be omitted.
Two main approaches for feedback prevention and control existed before Alango Quasi
Proportional Frequency Shifting (QPFS) technology was invented.

Automatic equalization
In this approach the signal spectrum is constantly analyzed in narrow frequency bands. If, on
any particular frequency f, a signal buildup is detected, this frequency is automatically
attenuated reducing D(f) . The problem with this approach is that the problematic frequency
may change constantly and it may be impossible to detect a “bad” frequency before it is too
late (the howling has already occurred).
Frequency shifting
In this approach the signal is modified so that its frequency spectrum is appeared shifted up or
down on certain amount of hertz. The output signal spectrum S

out

(f) is consequently given as

S

out

(f)=S

in

(f+

Df) . Frequency shifting is very efficient approach for treating the feedback

problem. It actually prevents the problem to occur because a microphone-speaker-microphone
loop on any particular frequency cannot be created as each frequency is constantly shifted up
or down. However, shifting in frequency violates harmonic character of some sounds. For
example, for voiced sounds in speech, all constituting frequencies are multiples of a
fundamental frequency called pitch. If the pitch frequency is f

p

than a voice sound consists of

frequencies f

p

, 2f

p

, 3f

p

, … nf

p

. Shifting these frequencies by

Df will give f

p

+

Df, 2f

p

+

Df,

3f

p

+

Df, … nf

p

+

Df . Obviously, the harmonic structure of such signal is violated since k(f

p

+

Df) ¹ kf

p

+

Df .

Human ear is very sensitive to harmonic structure so that such distortions are extremely
noticeable even for small values of the shift. In practice, frequency shift is limited by 3Hz,
which, in some circumstances, may be too small to prevent the feedback buildup.

background image

Acoustic Feedback Cancellation Alango Technologies Ltd. white paper

Page 4 of 5

Alango digital solution

Alango approach to the audio feedback problem is also digital where DSP is the central
element performing several tasks. Figure 4 illustrates the concept.

DSP

Howling

Preventor

(QPFS)

Subband

Howling

Blocker

Optional:

equalization,

Noise suppression,
Automatic Gain Control,
Multiband compression

Figure 4 DSP structure of Alango solution

The microphone signal is first transformed by Howling Preventor block. The general idea of
Howling Preventor block is based on frequency shifting preserving harmonic relationships
between frequencies. Ideally, Howling Preventor block would perform proportional frequency
shifting (scaling) where each frequency is shifted proportionally to its value. Unfortunately,
frequency scaling requires a certain processing delay exceeding the maximal allowed delay of
a “live” sound amplification system.
Howling Preventor block implements Alango proprietary, patent pending Quasi Proportional
Frequency Shifting (QPFS) technology. “Proportional Shifting” actually means scaling and
“Quasi” means that scaling is approximate such that S

out

(f)

@

S

in

(

a

f) . However, the difference

between the exact and “quasi” scaling is not noticeable due to a limited frequency resolution of
human ear. As a result, it introduces much smaller distortions by (quasi) preserving the
harmonic structure of the sound. With QPRS, the shift is not limited by distortions but by
apparent change in the speaker pitch. The concept is illustrated on Figure 5. The signal from
the loudspeaker to the microphone travels by air with its specific, linear filtering
characteristics. Attenuation may occur but the frequencies are preserved. On the way from the
microphone to the loudspeaker, proportional frequency shifting the output relative to input
breaks possible frequency loops. No feedback problem is consequently occur.

background image

Acoustic Feedback Cancellation Alango Technologies Ltd. white paper

Page 5 of 5

Df @ a f

F

QPFS

Figure 5 QPFS concept

Howling Preventor block allows for significant increase in the overall system gain
(amplification). However, when pushed to an extreme, the howling will still occur when the
amplification exceeds some level or the distance between the microphone and the loudspeaker
becomes very small. The objective of Howling Blocker is to detect such situation in a
particular frequency subband and reduce the gain in that band accordingly.
Optional processing block lists other Alango technologies that may be added without almost
any additional computational or memory requirements. Static equalization, noise suppression,
multiband dynamic equalization (compression) and automatic gain control are easily
implemented as a part of the feedback cancellation technology.

Technical specifications

Processing delay: 5ms;
Computational requirements: <5 MIPS;
RAM memory: <2KW

Demos

Real time demo of the technology is available using TMS320VC5510 evaluation board.

More information

Please, don’t hesitate to contact us regarding technical, licensing and other information:
E-mail:

info.sound@alango.com

Phone: +972 4 8580743
Visit our web site

www.alango.com

to gain more information about our technologies.


Wyszukiwarka

Podobne podstrony:
Rane Understanding Acoustic Feedback and Suppressors
Rane Understanding Acoustic Feedback & Suppressors
Pancharatnam A Study on the Computer Aided Acoustic Analysis of an Auditorium (CATT)
Acoustic Notes id 50875 Nieznany
PPw2 acoustics
Ouellette J Science and Art Converge in Concert Hall Acoustics
Egner Architectural Acoustics
Lynge Odeon A Design Tool For Auditorium Acoustics, Noise Control And Loudspeaker Systems
Articulation and acoustics Ladefoged and Johnson (2011; 8 17)
Iannace, Ianniello, Romano Room Acoustic Conditions Of Performers In An Old Opera House
Philips Motional Feedback Speaker
Gardner The Virtual Acoustic Room
24 CNC machine feedback devices
PRESENTATION FEEDBACK doc
feedback zastosowanie w pracy kierownika, nauka
FEEDBACK
Feedback na bieżąco w rozwoju kompetencji(1), kompetencje, zarzadzanie zasobami ludzkimi
Nijs L , de Vries The young architect’s guide to room acoustics

więcej podobnych podstron