Everything

FLAC vs MP3

A friend asked me a question on Facebook. Are FLAC files better than high bit rate MP3’s? I instinctually said yes, but I didn’t know how much better they were. So, I set out to find an objective way to answer this question.

I’ll start with my conclusion. MP3 files at very high bitrates add a surprising amount of distortion. What I found most interesting was that 320 kbps MP3 files weren’t much better than 192 kbps when compared to the control file. Or, maybe phrased more accurately, the distortions introduced in the files were much more pronounced than I expected, even at 320 kbps.

Methodology:

I wanted to start with the best possible source. I used a high-resolution FLAC file as my material. In this case, I opted for a complex song (Muse – Super Massive Black Hole) and a subtle song with wide dynamic range (Dan Wilson – Never Meant To Love You).

I used Adobe Audition to perform the conversions and analysis.

Both files were converted to 44.1 kHz sample rates and 16 bits per sample. I want to note that this conversion process alone introduces errors relative to the original files. This was a necessary step because MP3 files can’t contain high-resolution audio or 24-bit samples. By doing this step first, I would limit the variables later in the test.

From there, each file was saved as 192 kbps and 320 kbps MP3 versions with a sample rate of 44.1 kHz and 16-bit sample depth.

File sizes:

  • Complex FLAC – 44,693 KB
  • Complex 192 MP3 – 4,897 KB (88% reduction)
  • Complex 320 MP3 – 8,162 KB (80% reduction)
  • Simple FLAC – 34,431 KB
  • Simple 192 MP3 – 5,261 KB (85% reduction)
  • Simple 320 MP3 – 8,768 KB (75% reduction)

Comparison:

Each comparison was performed by aligning the original control FLAC file with the test file. I then inverted the control file and summed the result. This step cancels out the original music from the track and only leaves the differences and the changes. Below are the results of each test. The multitrack project was also set up to 44.1 kHz 16-bit samples. The default is a 32-bit float, but I did not want to introduce extra conversion steps in this process. All of the files have the same sample rate and bit depth.

You can download the difference file (compressed as FLAC) and hear exactly what noise and distortions were introduce for yourself. For each file, I’ve included a waveform, spectral frequency display, a snapshot of the FFT, and a snapshot of the phase plot. (The unmodified files are available for download)

Results:

Complex Song: Lossless vs. 192 MP3

Audio Differences


Complex Song: Lossless vs. 320 MP3

Audio Differences


Simple Song: Lossless vs. 192 MP3

Audio Differences


Simple Song: Lossless vs. 320 MP3

Audio Differences


Other reasons to use FLAC over MP3:

The primary reason to use FLAC over MP3 is that it is lossless. If you need to convert them to a new format in 10 years, you can do so without losing any quality.

If you like high-resolution audio – music with higher sample rates than 48,000 samples second or greater than 16-bit sample depth, then MP3 files simply don’t support what you need.

Lastly, storage is cheap. You can buy an 8 TB drive on Amazon for less than $200. Yes, the 320 kbps files take up 20% of the file space, but storage is so cheap, why would you even bother? And storage will only get cheaper. Once you have encoded the music in MP3, the errors introduced can never be removed.