loss of quality

yep. I have a vip account. originally, when I signed I wanted to upload a twenty one minute jam. I rarely post something that long now, but I enjoy the additional perks of the vip status.

Now sounds great even with my crappy speakers! :shock:

Just a couple of comments more.

The link is not correct, it is: http://www.soundclick.com/bands/page_music.cfm?bandID=644552
(Surprisingly the domain sondclick.com exist!)

You should convert from the original uncompressed format to MP3 at 320 kbps because if you convert from de 128 kbps you could not recover all the frequencies, they were already filtered.

doh!...Nuno, I corrected the address and added the 'u'.

and yes, I went to the original recording at reformatted to 320kbps.
another thing I did was to add foam under my studio monitors. the bass response I had been hearing is toned down some and probably better for when I mix.
little things matter when creating a whole.

another thing I did was to add foam under my studio monitors. the bass response I had been hearing is toned down some and probably better for when I mix. little things matter when creating a whole.

If you want to get more in depth on how the room can affect what you hear, check out some of Ethan Winer's articles on acoustics or watch some of his very entertaining educational videos (Warning: some of his videos use unconventional methods to illustrate his points. They're fun, but they're a bit risque.)

I like risque. thanks for the links.

Hey, Dogbite...

MP3 stinks. This is most often (but not always) due to what's called the Nyquist Frequency...

Cat

This is better applied to loss-less codecs, and high quality recording and reproduction. This is not that. The game is completely different for MP3, WMA, AAC and other lossy codecs.

Nyquist freq defines usable audio bandwidth in a sampled (usually digitized) signal => requisite low pass filtering to make sampling work without creating really nasty aliasing artifacts. But ... there are far worse things going on in a lossy codec than simple loss of high frequency content. As Moonrider mentions, the lossy codec is tossing out large (actually most) of the original information in favor of what "it believes" is most important to the listener and is within its capability to reasonably reproduce. In many cases, if one is limited to a certain bitrate lossy codec (say 128 kb/s MP3) it often pays bigtime to lower the pre-sampled audio bandwidth -- say from max of 20kHz down to 15kHz or even 10kHz, and even consider lowering the sampling rate to 22.05kS/s (Nyquist says it's possible at 10 kHz). This helps put bits to work in other areas of reproduction instead of bit-hogging high freqs -- which is a place many lossy codec will suck anyway (harshness, artifacts).

Remember that soundclick will automatically downgrade quality to 128kbps.

Ah, I didn't know that.

That explains a few things, certainly. Thanks.

But ... there are far worse things going on in a lossy codec than simple loss of high frequency content. As Moonrider mentions, the lossy codec is tossing out large (actually most) of the original information in favor of what "it believes" is most important to the listener and is within its capability to reasonably reproduce.

This is of course, technically true, but sounds much worse than it is. Most bits in music are either nothing, or sounds/frequencies you just can't hear. As we've said, when you're younger, you can hear more frequencies, and as you get older, you lose the very high ones (Not that old either, 25ish usually). Regardless of this, I visit a lot of forums where there are a lot of people who think they can hear the difference between FLAC (lossless compression) and 320kbps MP3's. I've challenged 3 of them that I know in real life, and a couple more people online who are honest enough to use the honor system, and I've never seen anyone who can tell the difference. I think if it's humanely possible to hear the difference, there is a microscopic section of the population who can (I still think that section doesn't exist). Now, to back up these claims, let me make a few clarifying statements:

1. Ripping is just as important (maybe more) than encoding. You need a 100% quality rip with no artifacts, skips, etc. This means using ripping software that will do multiple passes, and an unscratched CD. I used Exact Audio Copy (EAC) which is free, and awesome.

2. The encoding process should be CBR (not VBR). Constant Bit Rate, 320kbps.

3. The music needs to be listened to through high quality sound equipment. (If you don't have professional grade music listening equipment, 192kbps sounds the same as well.)

For practical improvements in home/hobby recording, there are additional, basic considerations that will make real differences, even if using modestly-priced equipment.

1. A multi-step lossy codec conversion such as lossless to 192 Mb/s to 128 Mb/s will generally result in worse audio quality than one conversion from lossless to 128 Mb/s. Applying this to multitrack recording: If your backing tracks or tracks supplied by another are already in lossy codec format, convert them to non-lossy format (something like WAV or Intel PCM files) for use in your recording/mixer app. Keep them that way through final mixdown and mastering, making a fully linear and lossless version at whatever high quality standard you usually prefer - CD/WAV/HDCD ... THEN convert to lossy MP3 or WMA only ONCE. For each bitrate MP3 or other lossy version needed, do a separate conversion from the fully linear, lossless version. So, if you are submitting your work to a site than will limit the bitrate, you should make a dedicated version in that bitrate directly from the high quality, lossless version.

2. Also keep in mind: If you supply your MP3 files to a site in their preferred bitrate, then they will not muck around with the stereo, bandwidth, bit-depth and other parms you may have carefully chosen to optimize your audio product. But if it has to do a conversion to lower bitrate, the site will simply apply a generic conversion in one-size-fits-all fashion. You lose.

3. Where possible, make the most out of your final bit rate. I agree that CBR generally works better than VBR. I've heard improvements result by lowering the bandwidth, mixing to reduce the stereo image (mono would be the extreme case) and sometimes dropping the sampling rate. Which measures to use will depend upon the type and content of the audio, and some consideration of the target playback system.

4. If one or more of your tracks happens only to be available in a marginal quality MP3, and seems to be full of annoying artifacts ... but you need it anyway, then there are several things you can try to mask that suboptimal quality. Push that track's level down a bit in the final mix. If it will fit the recording style, add some reverb to the track, as the time smearing of reverb can mask a lot of low level nastiness.
But ... there are far worse things going on in a lossy codec than simple loss of high frequency content. As Moonrider mentions, the lossy codec is tossing out large (actually most) of the original information in favor of what "it believes" is most important to the listener and is within its capability to reasonably reproduce.

This is of course, technically true, but sounds much worse than it is. Most bits in music are either nothing, or sounds/frequencies you just can't hear. As we've said, when you're younger, you can hear more frequencies, and as you get older, you lose the very high ones (Not that old either, 25ish usually). Regardless of this, I visit a lot of forums where there are a lot of people who think they can hear the difference between FLAC (lossless compression) and 320kbps MP3's. I've challenged 3 of them that I know in real life, and a couple more people online who are honest enough to use the honor system, and I've never seen anyone who can tell the difference. I think if it's humanely possible to hear the difference, there is a microscopic section of the population who can (I still think that section doesn't exist). Now, to back up these claims, let me make a few clarifying statements:

As part of my profession, I have managed and participated in rigorous double-blind lossy codec testing. And we set that up to rely on no honor whatsoever :wink: . One thing I'm sure you understand, is that there is at least one more critical variable in additions to the codec, its parameters (BW, sampling rate, bit depth, type of stereo encoding ....), listener and environment. There is the sound sample itself. The codecs are designed to work optimally with a certain psychoacoustic model of the typical listener (what a minefield) and with a statistical model of the content of the audio. When either of those deviates from the norm, performance collaspes quickly - more quickly at lower bitrates, of course. When evaluating a codec, we do not pick just a handful of genres of music, talk and sounds to test, but run through a lot of pretesting to select critical audio material that we find are good for eliciting artifacts. And these are not "tricky" samples, but 10 to 15 second snippets that come from real, listenable audio. In fact, one of the criteria is the test audio itself must be listenable and not contain uniquely distracting content, as that would compromise the testing. Sometimes it is very surprising what audio material will "drive a codec crazy." To these critical cuts are added less challenging samples, and then all are normalized for perceived loudness and put to the test using different classes of listeners: population, trained and expert. Listeners from the population are simply selected for demographics and reasonable hearing capablity. Trained listeners have been introduced to material that contains artifacts so they can learn to identify them from the original, high quality source audio. Training does not select people, but gives them a frame of reference. Training may or may not work for an individual. But value is significant for determining which artifacts become a permanent issue after being recognized once or twice, and which disappear back into the material as aural memory fades. Experts are not self-proclaimed golden ears, but persons tested and found to have a talent for detecting low level artifacts in one or more areas, such as imaging, HF, dymanics, freq balance, etc. They are essentially a set of specialized detectors, as few persons "hear it all." And the surpising thing is that (my guess) about 5% of the trained listeners turn out to have expert listening capabilities. I don't think that is a microscopic percentage. But that needs to be crossed with the distribution of codec-challenging audio material to understand how often listener will find the "codeced" audio objectionable. For MP3 across all listeners and program material, the stats are pretty low at 192 kb/s and vanishingly low at 320. But don't think that a few audio samples laid before a few self-proclaimed golden ears is an acid test. There is real world audio that will challenge MP3 at each of these rates. It just doesn't happen that often -- but then, that makes it a good biz model for bandwidth conservation.

A discussion of psychoacoustics and lossy codec design is pretty dry, but I will say this: what is thrown away in lossy codecs is not bits, but information. And the information considered significant is not necessary unmolested in transmission. Instead, the encoder may very well substitute a "sounds like" signal that is far easier to transmit at a lower bitrate. Sounds-like decisions and success all depends upon the psychacoustic masking that is estimated by the encoding analyses. It's not simple tossing of high freq info, because most listeners cannot perceive it. It's more like substituting a burst of filtered and time-gated white noise for a cymbal crash. All the encoder puts into the stream is "give me a noise burst at this time, duration, and use filter X1." The actual noise is never transmitted in the stream, as that's the part that requires a lot of BW. There are some parallels to MIDI and vocoding, though it's not exactly the same. But like MIDI, if the codec tool set (e.g. generator or patch set) isn't a good match for the source audio, the result can be very iffy.

BTW, being younger and able to hear higher frequencies does not make one a more critical listener anywhere except at high frequencies. Those that cannot hear the full spectrum also will not experience the full (desired) psychoacoustic masking of low and mid frequency energies that is crucial to making lossy codecs work. And thus, they often are more sensitive to lossy codec artifacts in low and mid frequencies. Try this: Filter a 128 kb/s MP3 with lots of HF content to remove those FHs above 4 kHs. Now listen to what's left, ignoring the dullness, but paying attention to warbles, imbalance, amusical noises ... Nasty, eh?