Friday, 30 September 2011

Playing music on your PC Part 1

This is my take on the whole FLAC vs mp3 controversy.

How do you check if a file is truly lossless?

From what I understand, one of the ways mp3 files are compressed from FLAC or WAV files is by cutting out the top frequencies in the spectrum. This can be seen on a Spectro* chart below. The top image is a FLAC version of the file. The peaks reach the top of the spectrum. The lower image is the same file converted to 320kbps mp3. The peaks in the second image look "chopped off". 

Track used was The Killers - Change Your Mind

Spectro is a useful tool for checking whether the file is a true lossless file or if it's a lossy file converted to lossless. The encoder used is displayed on the bottom left corner. You can get Spectro here.  It's free.

Anyway let's get back to the topic. According to the diagrams, lossless files have their high frequencies chopped off. In usual recordings the instrument that produces that high of a frequency is the cymbal. Unless you are very sensitive to high frequencies or have very good headphones and equipment that reproduce high frequencies very well, the difference between high bitrate lossy and lossless files are really tiny.

Downloading music from file sharing sites

Some of us are guilty of downloading music files through less than legal channels on the internet. Sometimes we notice a very significant between those and files obtained through legal channels. If both files have comparable bitrates, then there's a high chance that somewhere along the converting process the fellow accidentally introduced some noise into the file. I don't know how this happens but I have experienced this before. 

Blind Testing

Why? Because sighted listening is heavily biased. 

Test it out yourself
  1. Download Foobar2000. It's a great music player that uses a lot less power than iTunes. 
  2. Download the Foobar2000 ABX plugin here.
  3. Convert some lossless files (FLAC/WAV) to high bitrate lossy (MP3/AAC/OGG)
  4. Run the test.
Video tutorial (not mine)

If you can get a good result without guessing, give yourself a pat on the back.

Is a high bitrate mp3 really that important?

There's a phrase that is commonly said in audio forums; you need to have everything in 320kbps or FLAC if possible. If you have a very good headphone rig and you are the kind of person who does critical listening, then this is true. If you are that person, I don't know why you're reading my articles.

Anyways, with a very good system, the differences between a lossy and lossless sample is more apparent. But they skipped a huge step. Just having a lossless file does not ensure a high quality sounding track.  The mastering or recording of the track is more important than the bitrate of the file. Unfortunately, whether the mastering is good or bad, it's out of your hands. Mastering is entirely up to the engineers behind the album.

Some blow so much money into audiophile recordings only to find out that they're just fooling themselves. If you like listening to mainstream songs like Backstreet Boys,  Pitbull, etc, don't force yourself into buying audiophile tracks.

Building a system

This is how you should build your headphone rig, imo. Invest as much as you can into headphones. Use as much as you have left into an amplifier if you have power hungry and inefficient headphones (you can use stereo receivers instead), if not, get a decent DAC if you are still using your PC's onboard audio. Lastly spend some money on CDs or just download them off the internet and save the rest for a rainy day.

If you are on a tight budget, say.. $50, spend all of it on headphones. There are a couple of cheaper sub-100 headphones that are easy to drive and sound decent. Amplifiers come second. 

I find it hard to distinguish 400kbps AAC or 320kbps mp3 from FLAC but that doesn't stop me from loading lossless tracks on to my portable media player (Sansa Clip+). Sometimes it doesn't matter if the difference you hear between two things are real or imagined. If you think FLAC sounds better in a sighted listening test but you can't differentiate mp3 and FLAC in the Foobar2000 ABX test, then by all means do whatever makes you happy.