Solid state amplifiers and sound stage, especially front to back "depth"

No doubt different amps will have different transfer functions that could result in the parts of the signal that determine imaging and soundstage, many of which are very low level, being reproduced differently.
A linear transfer function is generally what is desired on paper but it’s likely the case that no two amps have exactly the same transfer function be it linear or not so linear. Transfer functions that emphasize the lower level parts of the signal that determine soundstage and imaging could even be artificially emphasized in a way that increases soundstage and imaging even if that is less linear. I would expect that amps that soft clip where dynamic range is compressed (many tube amps) and that also feature very low noise floors could excel in soundstage and imaging. Even if an artificial artifact of the amps non linear transfer function when soft clipping it might be regarded as a desirable effect.

In that context, my Bel Canto ref1000m amps also do soft clip but effectively never do in that they are 500 w/ch into 8 ohm doubling into 4 ohm. However, they are also absolutely dead quiet so together the effortless dynamics and low noise floor combine for quite exceptional soundstage and imaging (also dynamics) I would say with a very linear transfer function. Nothing artificial there.

The term "voicing" tends to carry over more from Guitar amp designers and modders looking for particular tones, overtones, and/or feedback designed into the circuit looking to achieve a particular sound result.

With professional and home audio, it seems many audio engineers focus primarily on specs and limiting distortion alone. Some others, changing component values, adding components, removing components, adding gain stages, or even adding complete circuits like negative feedback working to achieve a particular sound.  Agree or not, likely adding distortion in what they deem as "in the right places", also loosely referred to by some as voicing, fwiw. Results and opinions vary of course.    

A set-tube audio buddy likes to say, "at the end of the day, it's just your stereo", I'll add, call it whatever you want, you are the one listening to it. 

  

I guess I don't really know what that term can mean or how to interpret his comments

Well that makes two of us. If 'voicing' is trying to get the most linearity out of the circuit than we do that. But we're not listening to it when we do that- we're measuring it. The listening only comes in after it seems OK on the bench. So far the listening hasn't revealed anything we couldn't see on the bench in the last 45 years...

@decooney makes a good point with regards to guitar amps where you have to be careful to not allow the circuit to play bass and to use the proper cheap parts that give you that 'sound' when overdriven. Entirely different kettle of fish!

@atmasphere  I think I remember Paul McGowan saying they build the amp, then go play it in their IRS5 room and then go back to the bench to change it. In other words, they want they amp to "voice" in a way that they deem to be "musical," and those adjustments are not based on measurement but on listening. That's at least how I remember him explaining it. And that seems to be exactly what Frank was saying he doesn't do. But I remain confused. 

That's at least how I remember him explaining it. And that seems to be exactly what Frank was saying he doesn't do. But I remain confused.

@hilde45  All amps make distortion; if you understand how the ear interprets distortion than you don't have to do much if you understand how the topology of the circuit design affects distortion (for example, how single-ended circuits express a 2nd harmonic through a quadratic non-linearity). Put another way, this stuff is knowable and responds to math. If you did your math homework there won't be too much tweaking needed on the bench.