FLAC/WAV to MP3 (320 Kbps) --> 16 kHz cut-off

Hello Audacity users :slight_smile:

This “issue” has been going through my mind for several weeks so I finally decided to register on these forums and create a topic about it ^^

I was wondering as to why some of my music, which I convert from FLAC/WAV to 320 Kbps MP3 using Audacity 2.0.5 LAME, has a visible cut-off at 16 kHz, that is normally supposed to be at 20 kHz for 320 quality.

Here are 2 Spek screenshots to back up my claims:

https://i.imgur.com/gmVJwMg.png (it’s even more blatant on this one)

Any ideas? Thanks in advance.

Do you have to use LAME? Of course there are lossless formats or other lossy compression methods with different compromises.

has a visible cut-off at 16 kHz, that is normally supposed to be at 20 kHz for 320 quality.

I’m pretty sure the default is to cut-off the highest frequencies. And, it’s not designed to make nice looking spectrograms. It’s optimized for the best sound.

I think you can get around that using the [u]command line[/u] but I don’t know how to do it. It might be the -Y switch? There are several options that can affect frequency response.

I trust the LAME developers have chosen the best defaults. It’s lossy compression and if you force it to keep those highest frequencies something else will have to be thrown-away.

…Even if you can hear loud, pure, high-frequency test-tones in a hearing test in a soundproof booth, the highest frequencies in music tend to be weak and they get drowned-out by the other music. If you can hear a difference between the original and an MP3 on a [u]blind ABX test[/u] you’re probably not hearing the loss of high frequencies.

There may be other considerations as well. I created a timeline with pink noise and encoded it at 320 MP3. It carries all the frequencies and tones up to 19.7KHz as measured by Analyze > Plot Spectrum.

But I can well believe the stuff above 17KHz is getting whacked off because nobody is going to hear it anyway. If the main show frequencies and tones are loud and a major part of the show, preserving high frequency tones would be a complete waste of time.

MP3 doesn’t play by regular rules. Its mantra is “If you can’t hear it anyway, delete it.”


For me “convert from FLAC/WAV to 320 Kbps MP3 using Audacity” 2.2.1 has a 20kHz cut-off …

WAV to 320kps MP3, via Audacity 2-2-1, applies 20kHz cutoff, (not ''16kHz'').png

MP3 pays attention to content. That’s why it’s so hard to predict what it’s going to do.


I chose something with all the frequencies as a test : surf …

It’s lovely surf, but I show no energy above 11800.

When I export that to MP3, I show tones up to 14500…(!!) That’s MP3 rearranging tones for more efficient compression. That’s what gives you honky, talking into a wine glass vocal compression errors. That odd vocal tone is efficient to compress.


Yeah hence why I specified “SOME of my music”. The rest of my MP3s cut off at 20 khz just fine, as expected. There are only a few exceptions, such as those posted above. But thanks for your contribution.

That’s a lot of replies! Okay so according to you lads, getting spectrums like these when using LAME is absolutely normal and there’s nothing to be surprised about. LAME just straight up removes frequencies that are inaudible. Cool! :smiley: