I bought a record at HDtracks

I bought a record at HDtracks to see if it has any benefits in contrast to CD-like audio files at a sample rate of 44,1 kHz and a bit depth of 16. I want to share my findings with you.

I’m particularly curious about what a difference the sample rate makes. Concerning the bit depth, things are quite clear to me: It is proportional to the signal-to-noise ration (SNR): With 16 bit, it’s about 96.33 dB, which definitely is enough when it’s only about listening (I’m not talking about producing/mixing).

I chose the record “Tango In The Night” by Fleetwood Mac, as it was relatively cheap for 14,50€ and because I do know this album rather well. I know the songs and I know the fine transients of Lindsey Buckingham’s wonderful guitar picking in Big Love.

I had to go with the 96/24 version, as the 192/24 version was “not available in my territory”. Whatever. By the way, such high sample rates can make your music sound worse. Actually, 192/24 downloads are silly. “Its playback fidelity is slightly inferior to 16/44.1 or 16/48, and it takes up 6 times the space.” says the Xiph.org Foundation that produces audio formats and codecs like FLAC or Opus.

Bu my specific questions here were these:

  • Do I hear a difference between 96 kHz and 44,1 kHz?
  • Do I hear a difference between 48 kHz and 44,1 kHz?
  • Do I hear a difference with different sample rate settings of my audio interface?

So after downloading the record with the HDTracks download client, I converted the 96 kHz file to a 48k version and a 44,1k version. From the 44,1k version I converted a new 48k version. As conversion software, I used the invaluable XMedia Recode. I have converted all files to a bit depth of 24 bit. Distortion due to lack of dithering should definitely inaudible with 24 bit.

File List for comparison

Fun fact: Often people hear an actual audible difference between 44,1 and 48 due to the different behavior of their DAC converters at different sample rates.
Some audio hardware just works better/different at 48 that at 44,1. But that doesn’t mean that 44,1 is worse per se.

The playback chain

Steinberg UR44 + Beyerdynamic DT 770 PRO

My playback chain as follows:

  • Media player foobar2000 with ReplayGain set to off, volume to 100%.
  • All volume sliders in the Windows mixer set to 100%
  • It all goes then via USB 2.0 to the Audio Interface: A Steinberg UR44.
  • As headphones I’m using a Beyerdynamic DT 770 PRO (the 250 Ohm version)

I think this setup suits well enough for this kind of test.

The results

It was almost a bit disappointing, but my 27 year old ears haven’t heard any differences between any of the files. I conducted this experiment of course at several sample rate settings of the audio interface. There wasn’t any audible difference either.

I even did some blind tests (96k vs. 44,1k) with Foobar’s random playback function. Out of 5 times, 2 of my guesses were correct and 3 guesses were wrong.
I did this blind test with the interface set to 96 khz.

Pulling all those versions in a Samplitude project (with a sample rate of 96k, of course) and opening Voxengo SPAN, i was even more disappointed: Even the 96k version did not contain ANY frequency content above 20 kHz. In this case, the 96k version was completely needless. Thank you, HDTracks. Not.

SPAN Comparison 96k vs 44,1k

The (non-)lesson

Even though the HDTracks song is rather useless for comparison, I am still convinced that a sampling rate of 44,1 kHz (in theory) is just perfect for audio playback. Certainly for me and probably for most human beings including most of the people who call themselves an audiophile, because I have never met someone who could perceive sound above 20 kHz.

Considering the Nyquist-Shannon sampling theorem and the human ear’s abilities, 44,1 should suffice. After conducting their own tests, German consumer organization “Stiftung Warentest” also comes to the conclusion, that higher sample rates do only make sense, if you are a bat with the ability to hear ultra-sound.

There is this argument that you perceive music not only with your ears but with your whole body. That is true. You can feel the attack of a kick drum in your chest and a wobbly dubstep bass in your stomach. But don’t take it personally: Your body is too inert to resonate with ultra-sound. And if something like the hypersonic effect really exists is questionable.

There are of course reasons why to use 48 kHz when recording. I don’t want to get into that here.

Further reading

A lot of very clever people have dealt with this topic.

Justin Colletti says in a really elaborate and in-depth article:

It’s been proven again and again that even 44.1kHz can be completely transparent in all sorts of unbiased listening tests.

Monty at Xiph.org writes:

192kHz digital music files offer no benefits. They’re not quite neutral either; practical fidelity is slightly worse. The ultrasonics are a liability during playback.

Wikipedia refers to a study that says humans can hear sound as low as 12 Hz[6] and as high as 28 kHz. But the study admits that there could be measuring errors.

Recording and mixing genius Graham Cochrane always records and mixes in 44.1 kHz.

UPDATE 2016-05-05

Just discovered this video by Digital Audio mastermind Monty Montgomery. He says that there is virtually no audible difference between 44.1 and 48 kHz (at time 10:55):

 

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.