Hi. I have several question regarding the information that the Plot Spectrum shows.
When converting a 320 kbps high-quality MP3 song to a 256 kbps AAC, the Plot Spectrum analysis shows most (if not all) of the upper frequencies preserved.
However, if I convert it to lower bit rates, like 160 kbps, the analysis shows some upper frequencies (abt. 15KHz-20+KHz) being lost. Does this mean the trebles in the song are also eliminated, and thus, cannot be heard during music playback?
For a 320 kbps song that originally has 15KHz as its highest frequency, would it be practical to convert it to a 160 kbps than 256 kbps, since 160 kbps, as far as I have seen in the Plot Spectrum analysis, cuts/eliminates frequencies higher than 15KHz? (I plan to reconvert my music to lower bit rates for space)
Yes, but note that 15 kHz is very high in the audio frequency range and even for young people with ‘perfect’ hearing, sensitivity to frequencies is much reduced. In other words, you don’t hear much up there.
It’s difficult to say what settings will be “good enough” as that is highly subjective. What is “good enough” for one person may not be for another. In fact, when listening to two MP3 versions of the same music, people will frequently disagree about which sounds “best”.
Compressing audio data is full of compromises. There are lots of tweaks and customisations that can be made to achieve the “best” result for a given file size. Optimising the settings for a particular song can be very time consuming, and often not worth the effort, though there are some simple guidelines that are definitely worth considering.
Encoding to MP3 always reduces the sound quality.
If you start with a really high quality WAV file and encode it as 32 kbps MP3, the result will be little better than telephone quality.
If you then re-encode the MP3 as 320 kbps MP3, it will still sound like telephone quality. It will actually be a tiny bit worse than the MP3 that it was made from.
In order to get a high quality 320 kbps MP3, you would need to go back to your original WAV file and encode that.
The main trade-off with “lossy” formats is between file size and quality.
There is a limit to how small you can get the file before it starts to sound bad. The “kbps” is arguably the biggest single factor in determining the sound quality.
For music, all else being equal, “VBR” encoding will generally sound better than “CBR” encoding (Audacity supports both. The “Preset” settings are also “VBR”).
Mono music can be compressed more than stereo music before the sound quality noticeably deteriorates.
128 kbps is generally considered to be the lowest setting for reasonable quality stereo music, but if the file is mono you can get similar quality down to around 80 kbps.
These are both lossy formats. When you compress a file to 1/5th of it’s original size (or more) something gets lost.
These are psychoacoustic compression algorithms, which means they try to throw-away stuff you can’t hear. Even though you may be able to hear a 20kHz tone in a hearing test with a loud tone against a quiet background, you don’t normally hear the 20kHz overtones & harmonics in music. The psychoacoustic algorithms are throwing-away details that are masked (drowned-out) by other tones. If you are using high quality (high-bitrate) settings, you may not hear any difference between the original file and the lossy copy.
Of course, MP3 and AAC use different algorithms. I think AAC tends to preserve more high frequencies, and it’s supposed to be better. But, it’s only “better” if it sounds better… With most music, 256kbps AAC and 256 MP3 will both sound identical to the original (in a proper blind listening test) so we can’t say that one is better than the other, and we can’t say 320kbps is better.
We already know it’s lossy and some information is thrown-away, so looking at the spectrum doesn’t tell you much about the quality. Listen with your ears, not your eyes! You can tweak LAME to preserve the high frequencies, but then something more important usually gets thrown-away and it sounds worse.
Going from lossy-to-lossy is considered bad practice. We don’t always have a choice, but ideally you would go back to the uncompressed original and make the AAC file from that. If you re-compress an MP3 (even to the same bitrate) the quality loss may be noticable. AAC is supposed to re-compress without as much degradation. But, I’m not sure if MP3-to-AAC is better than MP3-to-MP3.
Apple say that AAC is better than other compressed formats.
Microsoft say that WMA is better than other compressed formats.
The Fraunhofer Institute say that MP3 is better, but only if you pay for it.
Lame say they make the only MP3 encoder that is still actively developed. Xiph say that Ogg Vorbis is best, and it’s the only one that is truly free and open source.
The last large scale double blind listening test studies that I have seen have put OGG, AAC, and MP3 (Lame) very close.