Loudspeaker compensation

AN3340

10/24

Doc ID 18453 Rev 1

6

Loudspeaker compensation

It should now be clear that using an amplifier with a flat frequency response in the band

20 Hz to 20 kHz does not guarantee a flat overall response for the listener. The global

performance is limited by the poor frequency response of the speakers.

To compensate for a non-ideal speaker response we need to modify the response of the

amplifier. This can be achieved using a set of independent tunable filters to modify the

amplifier response either over a bandwidth as large as possible or over a defined frequency

range. The filters could be analog filters in an analog system and could be digital filters in a

digital system.

family of amplifiers, have the advantage that the processor can also be programmed as

filters to tailor the amplifier response.

6.1

General procedure

modifying, step by step, the parameters of the filters. This manual procedure has, of course,

its drawbacks:

it is time-consuming to repeatedly modify each filter in turn and measure the frequency

response.

the result is user depended. The experience of the user to select the types of filters

(such as peak, low pass, high shelf) and their parameters (such as cut-off frequency,

gain, quality factor) plays a fundamental role.

6.2

Speaker Tune tool

The ST Speaker Tune tool available in the APWorkbench software allows very fast and

reliable frequency compensation of the whole audio chain and in detail the speaker

frequency response. The final result is unique because it is the result of an automatic

procedure based on an algorithm that manages the filter parameters in line with defined and

optimized procedures.

Speaker Tune is a very user friendly tool. It does not need special settings or specific

knowledge of filters. The only information that the tool needs is the frequency response of

the uncompensated loudspeaker.

The speaker frequency response can be imported into the tool in two different ways

graphically.

In the tabulated method, an example is given in Figure 23 on page 22, the first column

contains the frequencies and the second the sound level measurements. Maximum data

length is 4096 samples and maximum number of header lines is 3.

The frequency response can be measured using professional equipment and with an

anechoic chamber or using an MLS test (with microphone and equipment from Audio

Precision). The only stipulation is that the frequency response data must be expressed in dB

or dBSPL.