MP3s

[Giant Communist Robot]

Sampling is not the same as compression, and compression does not necessarily involve loss of data.

None of this argument is gonna fool me. A sample is not the entire population, just a representation--of the music in this case. And when the samples are replayed serially, you get a condensed, or shorted, or abreviated version. Or dare I say.......compressed.

[Shapley]

Sampling is not the same as compression, and compression does not necessarily involve loss of data.

None of this argument is gonna fool me. A sample is not the entire population, just a representation--of the music in this case. And when the samples are replayed serially, you get a condensed, or shorted, or abreviated version. Or dare I say.......compressed.

I recall some discussion back when the the CD format was first being released. Some audiophiles were concerned that the rate of sampling was insufficient for accurate reproduction of music. If I recall correctly, the stated reason for choosing the sampling rate was the requirement to fit one full hour (about the same as the capacity of a stardard 33 1/3 LP album, on a single disc. More samples would require more storage, which would require larger discs or more discs for the same amount of music.

V/R
Shapley

I think Barfle or Shos may have mentioned this back in an earlier thread, as well. My faltering memory leads me to believe I heard it before that, but I could well be mistaken.

[analog]

Wow. Nothing stays simple.
Not only is there range compression, which make soft passages louder and loud ones softer so kids can keep the volume all the way up, There's umpteen data compression schemes around now.
At its simplest, data compression might be replacing a moment of silence with an instruction telling the computer "next X milliseconds silent" which would free up the next X milliseconds of disc space for the subsequent music. That wouldn't lose any content.

MP3 is a scheme where somebody has decided what type of sound we humans won't miss, and the computer filters those parts out and doesn't write them into the file. Sounds shaky to me - surely all cochleas are not created equal and that's why Jamie says there's nuttin' like live.

From http://www.crutchfieldadvisor.com/ISEO-rgbtcspd/learningcenter/home/safemp3.html

Bigger files, better sound

There's a trade-off between file size and sound quality. Stuffing your MP3 player full of as many songs as possible can give you thousands of tunes to listen to, but at a cost. When you import a CD track, your music software program converts it to a sound file of a particular resolution, usually measured in kilobytes per second (Kbps). The lower the resolution, the smaller the resulting file — and the more detail lost.

Music with little detail tends to sound muddy. Most of us intuitively turn up such music in an effort to hear the missing details, and expose our ears to dangerously high volume levels in the process.

The software programs most MP3 players use, such as Apple's iTunes® or Windows Media Player 11, have default settings that determine what kind of file incoming music is converted to, and at what resolution.

Songs ripped from CDs are usually compressed to 128 kilobytes per second (Kbps) MP3 files. In the process of compressing the file, much of the audio information is discarded, leaving a general outline of the song. Much of the sonic detail that gives the music its character can either disappear or sound duller as a result.

This chart shows the relative file sizes of the same three-minute song saved in different formats, starting with the original CD track on the left. The smaller the file, the more sonic information is lost.



Anyone interested in the development of the CD might enjoy this IEEE article by one of the engineers involved:
http://www.itsoc.org/publications/nltr/itNL1207.pdf ---- page 42.

Sample rate and recording time were initially chosen at around 41KHZ and one hour to match up with digital tape recorders of the day on which their digital masters were recorded. That got changed - Page 44 has the famous ( and somewhat dubious) anecdote about how the playing time was finally decided by a Sony VP's wife, long enough to hold Beethoven's Ninth of which the longest known recorded performance was 74 minutes (Wilheim Furtwangler in 1951).

I just love trivia.

[BigJon@Work]

I started to study compression, sampling and loss about the time that single bit stream CD players started to appear to resolve the CD jitter errors that bothered some audiophiles. I found an article, in a science journal about the science of music and perception. What stuck out to me was that there are brief, sub-harmonic sounds that musical instruments make that we hear as uniquely identifying to the instrument. It’s part of how we tell a violin, from a viola from a cello, all playing the same note. My research at that time indicated that the current A to D converters of the time lost that sub-harmonic information and flattened the listening experience so that individual instruments were less realistic sounding. Id never pursued my research much beyond that point as I got bust with life and marriage.

[analog]

What stuck out to me was that there are brief, sub-harmonic sounds that musical instruments make that we hear as uniquely identifying to the instrument.

Are those the "overtones"(harmonics)?
I don't understand why it's not possible to capture and replay them. Maybe I just never had good enough hifi gear. But there is something about live music, some little tinge it has, that's instantly recognizable. Some people in my experience are indifferent to it, but i'd wager this community is keenly sensitive. I once walked a city block to investigate what turned out to be a guy on the sidewalk playing Beethoven's 9th on a Disston D23 handsaw. (He had good taste in saws.)

The ear does sort of a fourier transform - the nerve endings inside are each sensitive to a specific frequency. So I guess our brain receives the fundamental note and all the overtones as a simultaneous and rich pattern of impulses. I don't know why it likes a chord and dislikes discord, but sure it seems to be aware when the mathematical relation between the pure tones it hears is not an orderly one...

As an aside, it's overtones that are behind the "tube vs solid state" amplifier debate in the rock world. Modern solid state amplifiers, when overdriven into distortion as guitar guys are wont to do, add odd numbered harmonics to the signal. That's because they sharply clip both the top and bottom of the wave in a sharp and similar fashion, resulting in symmetric distortion, and symmetrical distortion of a sinewave adds odd harmonics - third, fifth, etc.
Tube circuits on the other hand drive into their upper and lower limits in a soggy and dissimilar fashion, clipping the wave tops and bottoms unequally with non-sharp and non-symmetric distortion... Non symmetric distortion of a sinewave makes even numbered harmonics - second, fourth etc. Non-sharp distortion favors the lower ones.

Apparently the ears of rock musicians have no trouble discerning the difference. And there's some psychological preference.

[BigJon@Work]

No it is not overtones, although they play a large part in distinguishing instruments. These are called sub harmonics because they never develop into a wave, they are created and disappear rapidly, too rapidly for the common early sampling techniques. It is described by the audiophiles who’ve heard it included, then excluded, from a recording as “taking the air” out of the instrument sounds. It is not a widely understood or researched topic and requires very high speed equipment to capture and analyze.

[dai bread]

Apparently the ears of rock musicians have no trouble discerning the difference.

I didn't know rock "musicians" had ears. Not after a year or two, anyway.

[barfle]

A sample is not the entire population, just a representation--of the music in this case. And when the samples are replayed serially, you get a condensed, or shorted, or abreviated version. Or dare I say.......compressed.

You might dare say it, but nobody who uses the term professionally would. And nobody's saying that a sample is the entire population, and nobody is saying that an analog recording is NOT just a representation.

[barfle]

These are called sub harmonics because they never develop into a wave, they are created and disappear rapidly, too rapidly for the common early sampling techniques. It is described by the audiophiles who’ve heard it included, then excluded, from a recording as “taking the air” out of the instrument sounds. It is not a widely understood or researched topic and requires very high speed equipment to capture and analyze.

And there are people who claim they can tell the difference between a $5 cable and a $500 cable.

[BigJon@Work]

And some can.

[analog]

No it is not overtones, although they play a large part in distinguishing instruments. These are called sub harmonics because they never develop into a wave, they are created and disappear rapidly, too rapidly for the common early sampling techniques. It is described by the audiophiles who’ve heard it included, then excluded, from a recording as “taking the air” out of the instrument sounds. It is not a widely understood or researched topic and requires very high speed equipment to capture and analyze.

Thanks for the clarification. My prior experience with signal analysis comes from industrial vibration and process signal measurement not music, so while the technique of capture and analyze is not unfamiliar the terminology is different. 'Subharmonic' to me would have meant an 'undertone', ie 1/2 or 1/3 the fundamental frequency.. what you describe i would have called a 'transient'.

We typically analyzed a time segment of signal to pick out what frequencies were present. To us 30 hz was a very high frequency. Short duration transients would be attenuated by a FFT program in the industrial gear of my experience - and maybe by a noise reduction algorithm in a recording studio's console?

This is speculation of course because I don't play an instrument-- but would you indulge me?
I can imagine applying high speed photography to a bow striking a violin string - it would send a wave down the string like a child's slinky toy and the bowstrings themselves would likely emit a tiny sound as they adjusted to the change in tension. Since the waves are not repetitive we industrial guys would call them transients. If the resulting soundwaves were high frequency short duration they might well be mistaken as digitization errors and ignored, at least by the equipment I'm accustomed to. If using a slow ADC they'd be purosely filtered out before sampling so as to avoid "aliasing " - a high frequency that's undersampled will be misinterpreted (aliased) as a subharmonic.

ADC's have got incredibly fast with the advent of digital video. Some enterprising audiophile marketeer will get the "air" back in. As (i think it was Barfle) mentioned the bottleneck now might be microphone technology.

Still that sidewalk guy playing the Disston crosscut saw had that "air" you mentioned and it caught my ear from afar. I don't know what it is.

Thanks for indulging me. I'll cease and desist now.

a.

[BigJon@Work]

Your undestanding tracks with mine. What we call a transient in my industry is a rapidly decaying wave, so for me that terminology would work either.