It has to be SPL vs. power spectrum differences. Even then, there still seems to be significant departure from personal experience.
I see more of the "pop spectrum" (figure 6) profile as being the closest to most pre-mastered performances without heavily stylized mastering practices, but with a more linear relationship in the 500-1000 Hz band, and typically more attenuation below 100 Hz--again, due to mastering practices to make tracks louder. This is the curve that I'd recommend most often for most design of non-commercial-application loudspeakers.
Most "loud" popular music, i.e., pop, rock with prominent electric guitar work, and derivative genres, seem to follow the "Heavy" profile (fig. 7) as mastered, but with more emphasis on bass relative to treble spectra power levels, i.e., a more downward tilt to the power spectrum curve.
Classical music seems to fit the Orchestral or Chamber profiles (figs. 3-4), but again with a more linearly attenuating curve above and below the peak power frequency in the 400-800 Hz region.
I note with no small amount of amusement that classical music isn't typically played at high SPL on sound reproduction systems that are subject to loudspeaker power management and design factors/margins. I wouldn't have included this genre in the typical power spectra for designing loudspeakers - rather, I'd pick something more typically stressing like the two genres mentioned just above. Perhaps the issue is with extremely low efficiency loudspeakers, like planars and very small surface area direct radiators playing classical, but even then, it strains the imagination.
Higher efficiency loudspeakers, such as horn loaded designs--the type more likely selected for commercial application--would not suffer nearly from power-related design issues while playing classical genres relative to the most stressing genres of dance/rock/pop.
YMMV.
Chris
I see more of the "pop spectrum" (figure 6) profile as being the closest to most pre-mastered performances without heavily stylized mastering practices, but with a more linear relationship in the 500-1000 Hz band, and typically more attenuation below 100 Hz--again, due to mastering practices to make tracks louder. This is the curve that I'd recommend most often for most design of non-commercial-application loudspeakers.
Most "loud" popular music, i.e., pop, rock with prominent electric guitar work, and derivative genres, seem to follow the "Heavy" profile (fig. 7) as mastered, but with more emphasis on bass relative to treble spectra power levels, i.e., a more downward tilt to the power spectrum curve.
Classical music seems to fit the Orchestral or Chamber profiles (figs. 3-4), but again with a more linearly attenuating curve above and below the peak power frequency in the 400-800 Hz region.
I note with no small amount of amusement that classical music isn't typically played at high SPL on sound reproduction systems that are subject to loudspeaker power management and design factors/margins. I wouldn't have included this genre in the typical power spectra for designing loudspeakers - rather, I'd pick something more typically stressing like the two genres mentioned just above. Perhaps the issue is with extremely low efficiency loudspeakers, like planars and very small surface area direct radiators playing classical, but even then, it strains the imagination.
Higher efficiency loudspeakers, such as horn loaded designs--the type more likely selected for commercial application--would not suffer nearly from power-related design issues while playing classical genres relative to the most stressing genres of dance/rock/pop.
YMMV.
Chris
I've found a strong statistically based -18 dB/decade (about -5.5 dB/octave) SPL downward slope trend in all forms of music - especially in "as recorded" (unprocessed) acoustic instrument recordings, versus "as mastered" music tracks processed for the mass marketplace (i.e., earbuds and mp3 players).
Power spectra shown in the figures would then need an additional -8 dB/decade slope on average to reflect that average. That doesn't seem to be reflected in either the IEC average music power level curve or in the B&O data, as noted above.
Chris
Power spectra shown in the figures would then need an additional -8 dB/decade slope on average to reflect that average. That doesn't seem to be reflected in either the IEC average music power level curve or in the B&O data, as noted above.
Chris
Last edited:
Hello
Maby they deleted it, the download was working after I upload it.
Bye
Gaetan
Hello
Here's a two new link to download the multitone test wav file, one of them should stay on line for few weeks:
http://www.megafileupload.com/e5mg/g_22hz_a_8khz_multitone.wav
filehosting.org | Download | g_22hz_a_8khz_multitone.wav
Bye
Gaetan
Here's a two new link to download the multitone test wav file, one of them should stay on line for few weeks:
http://www.megafileupload.com/e5mg/g_22hz_a_8khz_multitone.wav
filehosting.org | Download | g_22hz_a_8khz_multitone.wav
Bye
Gaetan
Last edited:
Coming across this ancient thread, the download links are broken, so here is a GNU Octave script to write your own file:
% want 22, 30, 52, 83, 123, 199, 316, 499, 802, 1248, 1998, 3152, 4997, 7997 hertz
fs = 48000; % sample rate [samples/s]
freq0 = 123;
freq1 = 199; % frequency of the tone [Hz]
freq2 = 316;
freq3 = 499;
freq4 = 802;
freq5 = 1248;
freq6 = 1998;
freq7 = 3152;
freq8 = 4997;
freq9 = 7997;
nsecs = 10; % number of seconds of the audio file
mtfile="mtfile.wav";
time = 0:1/fs:nsecs;
wave0 = (0.1*sin(time*2*pi*freq0));
wave1 = (0.1*sin(time*2*pi*freq1));
wave2 = (0.1*sin(time*2*pi*freq2));
wave3 = (0.1*sin(time*2*pi*freq3));
wave4 = (0.1*sin(time*2*pi*freq4));
wave5 = (0.1*sin(time*2*pi*freq5));
wave6 = (0.1*sin(time*2*pi*freq6));
wave7 = (0.1*sin(time*2*pi*freq7));
wave8 = (0.1*sin(time*2*pi*freq8));
wave9 = (0.1*sin(time*2*pi*freq9));
wave = wave0+wave1+wave2+wave3+wave4+wave5+wave6+wave7+wave8+wave9;
audiowrite(mtfile, wave, fs,'BitsPerSample',24)
% want 22, 30, 52, 83, 123, 199, 316, 499, 802, 1248, 1998, 3152, 4997, 7997 hertz
fs = 48000; % sample rate [samples/s]
freq0 = 123;
freq1 = 199; % frequency of the tone [Hz]
freq2 = 316;
freq3 = 499;
freq4 = 802;
freq5 = 1248;
freq6 = 1998;
freq7 = 3152;
freq8 = 4997;
freq9 = 7997;
nsecs = 10; % number of seconds of the audio file
mtfile="mtfile.wav";
time = 0:1/fs:nsecs;
wave0 = (0.1*sin(time*2*pi*freq0));
wave1 = (0.1*sin(time*2*pi*freq1));
wave2 = (0.1*sin(time*2*pi*freq2));
wave3 = (0.1*sin(time*2*pi*freq3));
wave4 = (0.1*sin(time*2*pi*freq4));
wave5 = (0.1*sin(time*2*pi*freq5));
wave6 = (0.1*sin(time*2*pi*freq6));
wave7 = (0.1*sin(time*2*pi*freq7));
wave8 = (0.1*sin(time*2*pi*freq8));
wave9 = (0.1*sin(time*2*pi*freq9));
wave = wave0+wave1+wave2+wave3+wave4+wave5+wave6+wave7+wave8+wave9;
audiowrite(mtfile, wave, fs,'BitsPerSample',24)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.