Recordings Through The Decades 2 - Frequency Range (again)
Part 2 of my article on how recordings have changed over the past 80 years. This time with sound samples.
This is part 2 in a series.
- Recording Through the Decades - Frequency Range
visual EQ comparisons of recordings through the decades.
- Recordings Through the Decades 2 - Frequency Range (again)
audio samples so you can hear the differences
Generated Sounds - Pink Noise and White Noise
There are two main kinds of random, generated noise used in testing recording equipment. One is White Noise, which is linear, meaning that at any frequency it has the same energy content. The other is Pink Noise (or 1/f noise) is logarithmic and has the same energy content per octave (doubling of frequency).
Pink noise is a random noise source characterized by a flat amplitude response per octave band of frequency (or any constant percentage bandwidth), i.e., it has equal energy, or constant power, per octave. Pink noise is created by passing white noise through a filter having a 3 dB/octave roll-off rate. See: White Noise discussion for details. Due to this roll-off, pink noise sounds less bright and richer in low frequencies than white noise. Since pink noise has the same energy in each 1/3-octave band, it is the preferred sound source for many acoustical measurements due to the critical band concept of human hearing.
( http://labs.google.com/glossary?q=pink+noise )
THIS GRAPH IS LINEAR while all the others are logarithmic. It shows visually how the white noise (shown in yellow) and pink noise (shown in red) are different. The following are two short sound samples of white and pink noise.
The theoretical ideal for most music would be an EQ curve resembling pink noise. This will sound the most natural to our ears, and probably represents what most music sounds like when performed live. In order to remove the music from the recording equation, what follows is EQ'd noise meant to match the recording qualities of 3 different eras of recorded music.
Noise Shaped To Match the Frequency Content of Different Recordings
Starting with the EQ curves of songs from 3 different eras of recorded music, I visually EQ'd some pink noise to match the recording. The pink noise is more compressed than the recording, but listening to these back to back, and looking at the graph should give you a very good idea of what the sound of recordings in each of these eras was like. You should hear it go from focusing on the mid range for vocals, go through the natural sounding recordings of the late 60's, and get more and more mechanical sounding towards our modern era, an almost complete inversion of what the earlier music sounded like.
Recordings from this era sound "thin" compared to our modern recordings, and emphasize the mid/vocal range.
The green line shows Pink noise, the red line shows Louis Armstrong's 1926 recording Cornet Chop Suey, and the yellow line is the results of some pink noise I EQ'd to resemble the frequency curve of the Louis Armstrong song.
Recordings from this era start to capture a fuller range of frequency, but still subtly emphasize the mid/vocal range. This era doesn't match pink noise exactly, but it's relatively close conformance to pink noise should make it pleasing to the ear.
Again, the green line represents pink noise, the red line represents the 1950's recording, and the yellow line is the noise I EQ'd to resemble the 1950's frequency content.
By now, recordings have pushed beyond the pink noise line, and emphasize strongly the bass range. These recordings should sound more like thunder, or heavy machinery than rain.
Lastly, here's Madonna's Ray of Light. Again, the green line is pink noise. The red line is the original song, and the blue line is the EQ'd pink noise designed to roughly match the frequency content of the song. The reason it drops off after 16 kHz is that I forgot to take the screen cap before I saved the file, so I reloaded the mp3 and used that.
Now you can both see and hear in a format totally unrelated to how you normally think of music - graphs and noise - how recordings have changed over the years. You can both see and hear how recordings have reversed their relationship to pink noise, which I feel reflects natural sound.
Pink noise occurs in nature, like when your television or radio is tuned to a dead channel, or perhaps like rain or the ocean surf.
You can draw your own conclusions by listening to the samples and looking at the graphs, but to my ears, the 1920's music sounds distant and wistful. Perhaps this is due to the Fletcher Munsen curve, and how bass and treble frequencies would fall off as a sound source got further from us. On the other hand, the 2000's sound reminds me of a jet plane taking off, or tons of water going over a waterfall or a movie where water is filling an airplane. It also sounds like industrial music.
In part 3, we'll deal with the volume and compression.
page first created on Wednesday, October 13, 2004
© Mark Wieczorek