i prefer to build something which measures well. I cannot comprehend why anyone would treat listening tests and measurements as an either-or divide. I do both.
What would I do if something "sounds good" but measures poorly? I would keep working. I would not accept something which merely "sounds good". It must measure well too.
In my limited experience, each and every piece of audio gear I have worked with or studied has sounded more "neutral" when it measured well. There are well-known examples of systems which sound good but measure poorly, e.g. significant quantities of lower-order even harmonics. I find such sound not to my taste -- it's like a cup of coffee with too much sugar. I prefer my single malt straight, no chocolate please. This is not a comment about those who like their sweeteners -- more power to them.
My last experience was with the Asawari Mark II. After I built it, the first few days were great, and then I began to notice problems. After listening tests and tweaking,
the sound became considerably more "flat", more neutral. It became more "ordinary" compared to the decidedly "hi-fi" character it had earlier. I realised that it was now more real-sounding. But this sound may not be to everyone's tastes.
I have studied in detail a lot of the blog posts from NwAvGuy, the designer of the O2 and ODAC. I have yet to find any divergence in our views (in spite of a million-mile gap in our knowledge levels
). He uses measurements and sets very high standards for measured performance. The outcome is an amazingly neutral sounding DAC and amp. I have purchased the O2+ODAC combo, and it is the most accurate and neutral sound reproduction component I have. I use it with my ER4P IEM -- this too is an IEM which is too neutral for some people's tastes.
On the topic of evaluating by ear: I think the ear is a wonderfully sensitive instrument, but it is also easily deceived. Listener expectancy is a very real problem. A lot of audiophile tweaks would never survive a proper double-blind listening test. I presume you all know about
this fantastic experiment? This does not mean that I recommend that you stop listening and go by measurement alone. In fact I recommend that you understand the (severe) limitations of both sorts of evaluation, be they instruments or be they your ears. Only then can you get good results out of them. For instance, some left-brained engineers believe that a single all-encompassing THD+N figure, expressed as a single percentage, is adequate to tell you how an amp will sound. This is immature -- you need a whole slew of measurements, with different loads and at different frequencies. For instance, see the set of measurements in this
review of a humdrum amplifier. The subtle differences in measured performance under different conditions tell their own story. In many cases, these differences make two amplifiers sound audibly different even when they have identical single-reading THD+N figures. I guess we all know how Peter Aczel evaluates power amplifiers by using a special device which applies different complex loads on it. People who thought that measurements began and ended with a single THD measure gave measurements a bad name, and that continues till today.
