Originally posted by colin traveller
And what is the difference between the above and this ...
For starters, the sample provided is a 128kbps encoded mp3 - not exactly ideal listening, for reasons enumerated below, but there is a difference between the frequency responses on No Boundaries and Goa nized. The bass response is louder and, while still more prominent at 100 Hz, there are fewer decibels between 100 Hz & 50 Hz. What you'll notice is that there's a notch of frequencies decimated predominantly after 16 Khz. Apart from the frequencies that are missing in action, however, there is more of a slope beginning at approximately 7,500 Hz. Your highs aren't nearly as prominent in Goa nized as they are in No Boundaries while your mids sustain a more prominent relation to the highs in the former than in the latter.
Goa nized
No Boundaries
Just listening without any analyzers, Goa nized is still harsh but nowhere near as harsh as No Boundaries. The harshness in the former, however, has more to do with how it was encoded than EQ practices. The low bit rate has only a minor role in eliminating "harsh" frequencies and best practices with your EQ should obviate the need to run a bit-crush version of filtering over the mix to give it a fictitious gloss.
quote:
Originally posted by colin traveller
i transfer it after recording at 320 is that 2 high or do you prefer that i should transfer at 192
The only thing that's really happening is that you're cutting detail out of the mix and actually making it less clear while introducing new and unwanted artifacts which constitute a harshness of their own. The saving grace of Goa nized wasn't that it was rendered to 128 kbps. It was that it was EQ'd substantially better than Music has no Boundaries.
The picture below is a mock-up of what happens to an audio signal that's encoded at particularly low bit rates. It won't be revealed in spectrum analysis but the phenomena is called aliasing and results in harshness when the computer is forced to split the difference between two bits because the original signal doesn't equal either of them.
The computer, not having an exact bit that equals the exact frequency of the original medium, dutifully duplicates the signal in two bits which are closest to corresponding with the original frequency. The lower the bit-rate, the more the computer is forced to make these arbitrary decisions, thereby resulting in inaccuracies that pollute the spectrum as they reduce its clarity. The same affect can be seen in low resolution photographs.
What's also happening is that you're essentially restricting the highest frequencies you can reach and while that might sound like it's a good idea, on paper, the problem is that those high frequencies, when they're not over-amplified, actually preserve detail in the mix that is sacrificed at lower sample rates.
In the interest of science, I went ahead and did some experimenting which confirms how restricted the frequencies at low kbps are. Unfortunately, I couldn't get down and dirty with your file in Wavosaur, but I was able to render a portion of it from SONAR into a new 96,000 Hz file that illustrates how badly the frequencies get cut.
First off, it should be noted that most people aren't ever going to hear frequencies higher than 18 kHz. Even when they hear them, they don't "notice" them in relation to the music they're listening to. Only the most anal retentive audiophile will admire how well the 17-22kHz range is filled out. Why are high digital audio frequencies important?
The Nyquist Theorem states that for every two Hz of digital spectrum, one Hz of actual audio can be replicated in the digital domain. If I am limited to 10 kHz of digital audio, the highest frequency I can duplicate, then, is 5 kHz. Since all mp3's render to 48 kHz of Digital Audio, the highest audio frequency attainable is theoretically 24 kHz.
However, because frequencies are also constricted by the frequency range of the original .wav file and the kbps of the mp3, not to mention the limitations of the encoding software being used, I've rarely seen them exceed 20 kHz. You can actually tell when someone has rendered a 320 kbps mp3 from a 44 kHz wave file because the mp3 will drop frequencies after 17,500 Hz, approximately.
...in case you're interested in hearing what I'm referencing. What's coming out of the player is 128 kbps, but the download is 320 kbps - shitty mixing and all. I'm putting it up because it's instructive, particularly in relation to your mix, rendered at 96,000 Hz, to understand what an mp3 file does to the bandwidth.
Goa nized @ 96 kHz rendering:
Notice the destruction after 16 kHz. Minus some anomalous peaks and the noise-floor, there's really not much there.
Here's the 320 kbps version of my song:
And here's the 128 kbps version of the same song:
Now, personally, I don't really mind listening to DJ mixes rendered to 192kbps. It's sufficient to discern the detail and easy enough to listen to. The problem I have is when you're considering lowering the quality to account for a mixing error. What's more is that it's not going to correct the error and it would probably have made the high-end problems even more apparent because there would be more persistent aliasing in that part of the spectrum. It just seems to be cutting off your nose to spite your face, throwing the baby out with the bath-water, and shooting yourself in the foot, all at the same time.
