DSD-Guide.com

Direct Stream Digital Audio

Believe What You Hear, Not What You Read. DSD - WAV - FLAC Conversions - What Does It Mean?

Believe What You Hear, Not What You Read. DSD - WAV - FLAC Conversions - What Does It Mean?

There has been a lot of controversy over the years about whether WAV conversions to FLAC are exact duplicates and sound exactly the same.  It's Ears vs Math.  Which do you subscribe to?  The Crew at Blue Coast Records believes there is a difference after doing dozens of blindfold listening tests over the last decade.  Why is this important to DSD listeners?  Understanding how to do a blindfold listening test is complicated.  Included in this article are methods how you can have a blindfold listening party to test for your own ears.

Written by Cookie Marenco, founder, producer, chief engineer at Blue Coast Records 

Since 2009 we have conducted dozens, if not hundreds, of blindfold listening tests.  Our tests have at least 3 people present, one to operate the devices and two to listen without talking to each other.  The operator does not tell us what they are playing.  

We usually listen to the same 15-30 second sample second of the converted file or using the same file with different devices.  It requires utmost concentration, quiet, stillness of motion and a great quality recording where you know the source is accurate.  A good set of headphones or speakers helps, but often, once you learn what to listen for, you can do the test on less expensive gear.

In an upcoming article, we'll explore more precisely what to listen for.

Very important... When you do the comparison test, whether in headphones (easiest) or good speakers, remember to remain still in your body position because every movement of your head will affect the appearance of sound (wall reflections, etc).

Many people believe that WAV converted to FLAC files should be exactly the same.  In 2008, we conducted the first comparison test before launching our digital download store.  Sure, FLAC files would be more convenient for everyone and save us money for storage and download bandwidth, but before we made that decision to stand behind the theory of "sounds exactly the same" we wanted to prove it to ourselves and these were the best files we could offer our customers.

Unfortunately, we found that not to be true.  In blindfold test, we could tell the difference.  The original file always sounded better -- more open and a wider stereo image.  We tested various levels of FLAC conversions and found that they all performed differently as well.  Even reconstituting the FLAC back to WAV(2) did not result in the WAV(2) sounding like the original.  So, we made the decision to release WAV files in 2008.

As we grew and released DSD for downloading, we continued to test the various conversions and methods... finally designing a proprietary method that we feel delivers the best sounding files.   We're not opposed to convenience.  We now sell FLAC files if our customers what those files.

We've tested various converters and found no two deliver the same sounding audio.  What we look for (and found) in various converters of DSD, WAV to FLAC were slight differences in how left and right balances changes, very slight collapsing of the stereo image, very slight phase changes at certain frequencies.  Granted, we do this for a living and the kinds of changes we are looking for are very minor. They require exacting listening test and are not probably perceived on first listening.  

We chose the converters that sounded best to us and best replicated the original master sound.  It's part of our proprietary system for conversion so I can't disclose how we do our conversions, but I can say that we've downloaded files from competitors sources and are able to recognize which converters were probably used on music we were all selling (that came from the same source).

The original file as sent by the record label will always sound better than any conversion.  At this time in history, a copy of the original digital file made from a converter will sound different than the original. If a WAV was converted to another WAV it would not sound as good as the original.  We have been doing this test since the mid 80's.  

We complained to Digidesign back in 1989 about "bouncing" tracks when we first heard the differences. Protools was selling the idea that there were no differences.  I had a friend on the inside who brought my test to the engineering team and 3 years later confirmed my suspicions.  Since then, I've been skeptical of any company saying "no differences in conversions".  Now we do the tests in house before making a judgement.

The concept is -- if we can perceive a difference in one small section, that is enough to confirm it is not an exact copy. If two of three people can describe the same difference in a blindfold test, then we can't sell the files as exact copies.  At our sister site, Downloads NOW!,  each MicroStore offers the provenance of every recording so that the customer can decide what works best in their system.

If you've done the listening tests and it makes no difference to you, then you are blessed not to worry about it.  We're not trying to persuade anyone to change their listening habits.  But, if you put your reputation on the line and sell products to passionate fans, then you need to do the listening tests to the best of your ability.  That's what we do.