Frequency Response: A Show and Tell Page 2
What does it matter?
Should you care if all the tones come out about the same? Absolutely. Generally speaking, you want to hear the music or the movie soundtrack in a way similar to what the artists and producers heard in the studio or on the dubbing stage. If your speaker sounds overly bassy, Nat King Cole's version of "Mona Lisa" might start to take on the booming tone of Lil Wayne's "Project Bitch." Far be it from me to speculate as to whether Cole would have approved of such a sonic transformation, but if the music you're listening to is going to be altered, the decision should be made by you, not by your speakers.
How much frequency response errors matter depends on:
1) The magnitude of the error (i.e., how big of a boost or cut it produces)
2) The frequency of the error (i.e., where it is in the audio spectrum)
3) The bandwidth of the error (i.e., how much of the audio spectrum is affected)
Magnitude of the error
Let's devote as little space as possible to this completely obvious rule: The bigger the frequency response error, the more likely you are to hear it. In other words, a +12 dB boost is more audible than a +6 dB boost.
Frequency of the error
Frequency response errors are more apparent if they occur where your hearing is the most sensitive and where even the worst audio system has an appreciable amount of response: in the midrange. Take this snippet of a tune from NYC-based folk/rock singer/songwriter George Stass, which I post-processed to demonstrate these concepts. You hear the 13-second snippet with the sound as recorded. Then you near the same snippet with the bass rolled off by -6 dB at 20 Hz (figure 2). Then you hear it with the treble rolled off by -6 dB at 20 kHz (figure 3). Finally, you hear it with a -6 dB cut centered at 2 kHz (figure 4).
The snippets with the bass and treble rolloff should sound pretty good. Depending on your hearing acuity and on the quality of the system you're using, you may or may not hear a difference. Either way, the sound won't suck. But the last snippet - the one with the midrange notched out - will sound obviously bad.
Same amount of cut, in three different places. But very different results.