I recently tried a little test to help educate myself. The result wasn’t at all what I had expected but I can’t think of any reason why.
I ripped a song from a CD three times - once as a WAV file, once as a 320 kbps MP3, and once as a 128 kbps MP3. I loaded them all into Audacity and wanted to see the differences in the waveforms. I expected the WAV to be most dense and the 128 kbps MP3 to be the least dense. Superficially, they appeared the same so I started to zoomed in to see how they differed. I zoomed in until I could see individual points and the waveforms all still looked identical.
The WAV file is supposed to be lossless and the MP3 files are lossy (lower bitrate = more lossy) is my understanding. As such, should I not see differences as I zoom in? What am I missing here?
320 MP3 and 128 MP3 are for all practical purposes perfect formats. 128 used to be the Audacity default MP3 setting. You can’t tell what happened by listening.
If you want to see the differences, open the 128 work in Audacity and export it again at 128. that should give you a reduced quality file and the damage should be visible. That’s the real problem with MP3. It’s a time bomb. You might not see the damage right this second, but next week, when you want to use the work in production for something else, you lose.
There was a poster who created a radio show by downloading MP3 songs and then critiquing them. He created an MP3 submission for the station and he might have gotten away with it in broadcast, but the station tried to make an MP3 podcast and couldn’t because the musical sound quality turned to garbage.
If your work is stereo, export it as 64 quality. That’s the MP3 quality that many people start to notice sound damage.
You might note that ACX requires MP3 audiobook submissions at 192 minimum, with mono strongly suggested.That’s a business decision. It’s high enough quality for them to make other products and services, but yet compressed for more efficient storage. You should not do that. You should store all your work in WAV, save it as archive and make the MP3 when you need it with appropriate compression.
WAV does have a quality value. I think the numbers come out to about 1411 for stereo, 44100,16-bit.
If you have the original WAV, and make an MP3, it is possible to isolate the differences :
the artifacts created by the MP3 encoder, see …https://soundcloud.com/magwhyr/modernist
It’s best to judge sound quality by listening, not by looking at waveforms. But, you’ll probably see a difference if you look at the [u]spectrogram[/u], especially if you compare the spectrogram of the uncompressed and compressed files. (You may have to change the spectrogram defaults so you can see up to ~20kHz.)
Also, if you run the Amplify effect it will indirectly give you the peak values - For example, if Amplify defaults to +1dB, you have 1dB of headroom (your highest current peak is -1dB). If Amplify defaults to 0dB, your file is normalized for 0dB peaks, etc. (You can cancel the Amplify effect if you don’t want to make changes.)
If you make an MP3 from an uncompressed original, re-import it, and check Amplify again, the highest peaks will often be higher (without changing the perceived volume). The wave shape changes making some peaks higher and some lower, so overall the new highest-peaks are usually higher. This is especially true with commercial music which is limited & compressed.
If you have the original WAV, and make an MP3, it is possible to isolate the differences :
the artifacts created by the MP3 encoder, see …> https://soundcloud.com/magwhyr/modernist
I assume that was done by subtraction. You can do that in Audacity by inverting one file and mixing.
Subtraction DOES give you the difference and if there is no difference you’ll get silence. But it’s important to know that “the sound of the difference” is not the same as “the difference in the sound”. For example, you can delay a copy by a few milliseconds (by adding silence to the beginning) and that doesn’t change the sound at all. But subtraction will reveal a huge-loud difference. The difference doesn’t sound like a delay (as long as it’s only a few milliseconds) but experts may recognize the comb-filtered sound as the result of combining (or subtracting) a delayed copy from the original.
MP3 tries to throw-away the “little details” that are masked (drowned-out) by the louder sounds. When you look at a waveform you are unlikely to see those little details. Plus it uses a couple of other “compression tricks” and some of those methods non-lossy.
([u]Inside the MP3 CODEC[/u])
But, you’ll probably see a difference if you look at the