In student times, an inquiring mind did not give rest. The question “who is longer and fatter?” Made me look again and again for adventures on my head.
What is better? wav, mp3 or ogg? I think that an ordinary habraiser is able to set priorities on its own. However, I am almost certain that there will be those who are not aware of why one format is better than the other, and the other is worse than the first. In this review, I want to share the results of our modest experiments conducted by my good friends some time ago. As a result, not the latest software versions were used, but I think the overall picture is quite clear.
Experimental setup
')
Oscilloscope: ... 1 pc.
Pulse generator: ... 1 pc.
Personal computer: ... 1 pc.
The purpose of the experiment
Show visually how the waveform changes, after being processed by various formats.
Experiment
The basis of the research was spectral analysis, i.e. frequency components in each of the presented sound files. Available software allowed to simulate the ideal sawtooth signal with a frequency of 333 Hz and an amplitude of 500 mV, the spectral composition of which is favorable for analysis. These monstrous experiments were put on him ...
First, let's see how the ideal signals look like:
The perfect 333Hz 44100 sawtooth
Perfect 333 44500 wav RF spectrum
Now look at the real signal. The real signal is the perfect one lost by Winamp.
Real Sawtooth 333Hz 44100 (WAV)
Real 333 44500 wav RF spectrum
For those who are not very well oriented in oscillograms or are no longer oriented at all, note: the straighter the lines, the better. And in the spectrograms a little differently: the more the graph looks like a real WAV (you never hear the ideal anywhere), the better.
We will consider the high-frequency region (13–23 kHz), since there is nothing to consider in the rest of the region, there is a very weak unevenness. Further. We clamp perfect and real signals from WAV to MP3 using Lame.dll and CDEX program.
MP3 Lame395mmx rel 44100 192kbps q = 0
CDEX mod1 parameters
CDEX mod2 parameters
CDEX mod3 parameters
Mod1 - selected from the recommendations of the manual to the Panasonic CT-720. Frankly speaking, weakly in both cases, a strong fall (4 times) after 17 kHz, the spectrum is thin.
Mod2 - to find out what q = 0 is, not q = 5. It affects very little, the spectrum does not change much, but becomes more even, the blockage remains.
Mod3 - Selected because I like the sound at 255kbps, I decided to see why ... The result is obvious! Density and uniformity increase sharply, the blockage moves to the region after 20 kHz and decreases by half. For ease of comparison, immediately after it again stands the spectrum of real WAV. The similarity is directly related! ..
We can also add that at 192kbps the uniformity increases and the blockage shifts to 19.2kHz, and at 320kbps everything disappears altogether, the density is slightly less.
Slides
perfect signal compressed MP3 Lame397b2 44100 192kbps q = 0
perfect signal, 333
44500 mod1 mp3
perfect signal, 333
44500 mod2 mp3
perfect signal, 333
44500 mod3 mp3
perfect 333 44 500 wav RF spectrum
perfect signal compressed MP3 Lame40A14 44100 192kbps q = 0
Real Signal, 333
44500 mod1 mp3
Real Signal, 333
44500 mod2 mp3
Real Signal, 333
44500 mod3 mp3
Next are the results of testing the ogg format, with which the same manipulations were carried out, for this I propose to become familiar with beautiful graphs, without unnecessary text.
wav 333Hz 44100
Real 333 44500 wav RF spectrum
Ogg 4 128
Ogg 6 196
Notice how much OGG is different from WAV and the MP3 considered earlier.
Spectrogram:
Real 333 44500 OGG 192
Acknowledgments
Many thanks to the user JIN: the hero of the soldering iron and the oscilloscope, who did the described experiments. Thank you, Eugene!