"...nothing gets you closer to the music."

Omega Mikro​

First, the bottom line: When a component gives you the the option of connecting balanced or unbalanced the sound in all cases (we and our customers have so far) tested (certainly, not an exhaustive study, but very highly respected components e.g., D'Agostino Momentum amps) was better unbalanced.  

The theoretical and engineering attraction of balanced topologies is that they suppress even order, especially second order, distortion products. That's a good thing*. Balanced topologies also are less likely to produce hum when there are ground loops and they can cancel  and reject some types of electromagnetic interference, also a good thing (especially if you run a recording studio or have a very complex system consisting of many interconnected components all powered by the local power company).

So why does the unbalanced mode, when both are available, sound better than balanced? My experience as a radio frequency (rf) engineer  suggests that what is probably going on in audio and rf are similar but with different consequences.  In rf, the  harmonic and intermodulation distortion are actually harmful because these distortion products can degrade communication links.  The case for audio is a little more complex and somewhat  contradictory.

intermodulation and harmonic distortion (of low order) are built-into our hearing mechanisms.  Research has shown that  our ears are mini speakers in that they produce intermod and harmonic distortion products when sound enters. (Our ear doctors measure these distortion products as a test of hearing quality).  So a little low order distortion in our music is not terribly harmful to our enjoyment of the music since our own ears are contributing a little of their own distortion to the music.. But as the distortion products become more complex and of higher order (say 7th, 11th, 23rd or 121st order products) they become more disconnected from our hearing mechanisms and the music and, as you might expect, annoy us 

In a former career, i worked in a lab where we characterized high power linear rf amplifier topologies for their distortion performance and found that balanced topologies did have the lowest second harmonic distortion products. But we also found that their higher order distortion was worse than their unbalanced counterparts. i think audio  and rf  amps share a similar quality (albeit for different reasons) in that higher order distortion products are more harmful to the mission than lower order products.

Now for some techtalk.  In ideal balanced topologies,  say a push-pull amplifier, identical active and passive components work together so that they cancel undesired even harmonic products while at the same time adding the desired signals. If both halves of the push-pull circuit are identical this works. But, in the real world, no two components are identical and in the push -pull process the extra non-ideal components needed to satisfy the balanced topology add together to increase the higher order distortion products** instead of cancelling them.  Unbalanced topologies can do the job with fewer components.

Balanced vs Unbalanced ----- which is better?

* you would think. However, several designers are purposely adding 2nd or 3rd order harmonic distortion to their amp designs because they say their customers, and they, like the sound better with distortion.

** Some might say that negative feedback will reduce those products so why worry? The problem with negative feedback is that it works well for steady state signals but does a poor job with transient signals like music.  So measuring a component's harmonic or intermodulation distortion using the standard sets of steady state tones  tells us nothing about how that component will sound while processing music sIgnals (which are never  in steady state).