We see in the gradual evolution of Federation starship design from the Constitution Class to the Intrepid Class a much more aerodynamic design in starships. We know that the Intrepid class has landing gear and that warp drive does NOT mandate an aerodynamic shape (ie Borg Cube), so is it reasonable to assume that Federation starships by the time of VOY were being fitted out with landing gear, hence the aerodynamic designs (like that of the Sovereign class)?
While this question appears to be opinion based, there are some things easy to spot if you pay close attention and think a bit about it. So it's not pure speculation.
Let's have a look at one specific scene during Voyager's opening credits:
As you can see, the gas clouds seems to react to the deflector field and/or shields (by lighting up), but at the same time that happens pretty close to the actual hull.
So even if you assume that no "space particle" ever touches the hull, a streamlined design might still be advantageous simply due to the fact that this makes it easier to project a field around the ship that is streamlined as well.
If you've got the technology, you can most likely adjust the field to ignore your ship's shape and/or simply overcome these issues as a whole (Borg cubes).
If you want a real world experiment or analogy: Think about how you could use garden sprinklers to create a "force field" of wetness. If you place them around the shape of your "ship", the resulting field will have a similar shape. If you don't want that, you'll have to increase pressure (i.e. more force/power/energy), which isn't necessarily that ideal if you can just (re-)shape your ship to compensate for that.
Now one might argue that in most combat scenes, shields appear as spherical bubbles around ships. This is true, but IMO this is just due to two things:
It's far easier (cheaper and faster!) to render; especially considering that rendering tech was very expensive (and slow) in the 90's.
It's a common misconception regarding "starship design". A sphere is a perfect shape and what you'll get from one single emitter (if it isn't directed/shielded in some way). But as all Star Trek series suggest, there's never just a single shield emitter. Most ships should feature at least four emitters (otherwise it wouldn't be possible to have shields collapse at one point of the hull only for example).
The Star Trek: The Next Generation Technical Manual, which was written by Rick Sternbach and Michael Okuda who were visual designers for the show as well as the main "technical consultants" (on some combination of real science and the reasonably consistent fictional science of the show), suggests that shape of the Enterprise-D was determined largely by warp physics, as well as the new requirement that the saucer be able to separate and from the "Battle Section" and coast at warp for some time (see Tango's answer to this question for more on this idea of the saucer "coasting" despite lacking its own warp engines). From p. 5:
Seen from a comfortable distance of two or three kilometers, the starship takes on the graceful lines of a nonrepresentational organic sculpture. Nature has determined the flow of the design, adhering closely to mathematical formulae at work in the universe surrounding the builders. Even in the desire to expand beyond the apparent limits of the natural world, familiar forces create familiar shapes. As the rapid aquatic and avian creatures of dozens of habitable worlds independently developed the unmistakable attributes of streamlining, so too did their interstellar cousin.
The combination of forces produced within the warp engine core and the flow of space and subspace around the vessel created the particular engineering solution to the problem of faster-than-light travel. The initial Starfleet requirement that a single spacecraft be able to perform as three distinct vehicles presented some rather complex—though some engineers not normally afraid of numbers preferred the word "daunting"—computational challenges.
The docked configuration presented the most efficient use of warp flight forces, but the Battle Station was also required to perform the specifications at warp velocities on its own, and the Saucer Module would have to have the capability of high sublight speed and possibly survive a separation at high warp. Scientists and engineers throughout the Federation, with all the deportment of composers and conductors, arranged sweeping curves, described vast volumes, and summoned up fantastic energies to bring their creation into existence.
Also on p. 66 they write:
Third, the shape of the starship hull facilitates slippage into warp and imparts a geometric correction vector. The Saucer Module, which retains its characteristic shape from the original concept of an emergency landing craft, helps shape the forward field component through the use of a 55º elliptical hull planform, found to produce superior peak transitional efficiency. The aft hull undercut allows for varying degrees of field flow attachment, effectively preventing pinwheeling, owing to the placement of the nacelles off the vehicle Y-axis center of mass.
Also, in the book U.S.S. Enterprise: Owner's Workshop Manual, written by Ben Robinson and Marcus Riley with Michael Okuda as a technical consultant, there's a sidebar on p. 97 titled "Warp Development Theory":
The warp "bubble" in which a starship rides at faster-than-light speeds is not a static structure. Rather, it is a flowing, pulsing, ever-shifting field of energy that dictates the shape of a particular starship. Even minor changes in starship design must undergo extensive testing and optimization to ensure safety and propulsive efficiency.
Field geometry is particularly sensitive to variations in the position of the warp engine nacelles. While in conventional (slower-than-light) spacecraft design, the propulsive thrust vector must travel through the vehicle's center of mass, warp propulsion is quite different. Warp drive requires the nacelle axes to be offset from the vehicle's centerline to create the propulsive imbalance. This technique usually requires the forward field lobe (normally containing the saucer section) also to be offset from the nacelle vector. In many older ships the nacelles were located slightly above the forward lobe. The NX and Constitution-class starships used this design, which yielded superior field stability, although it also resulted in reduced propulsive efficiency.
During the 2260s, the Advanced Starship Design Bureau began to experiment with field geometries that lowered the nacelle vector with respect to the forward field lobe. This approach promised improved engine performance but required several years of research into new field-stability software. Among the first vehicles to employ the new approach was the experimental Starship Excelsior, whose design lowered the nacelle centerline so it was level with the saucer module. This design was reflected in the Excelsior-class Enterprise-B. Further advances in field control software allowed even more radical designs for the Ambassador-class Enterprise-C, and the Galaxy-class Enterprise-D, both of which saw the nacelles move even lower than the frontal lobe.
Strangely enough, recent advances in deflector field technology have resulted in a reversal of this trend in nacelle position in some of the latest ships. New ultra high frequency shield generators have allowed the Sovereign-class Enterprise-E to employ higher nacelle positions, while enjoying an additional 3 per cent in engine performance.
The arcane engineering of warp field geometry continues to be a subject of intense study at Starfleet's Advanced Starship Design Bureau and will undoubtedly continue to be an area of ever-changing research and innovation.
I think the word streamlining would be a better fit then aerodynamics. Even Voyager used raw power over principles of lift when in atmosphere. 1701 D's saucer section was only designed to land as a last ditch effort and not as routine. Moving through space though, does present problems similar to atmospheric flight. Space is littered with micro meteorites and other debris. A minimal frontal exposer, helps to reduce the risk of collision damage. Most designs also include a form of navigational deflector array just to prevent damage. For protection in situation such as combat, another system of shielding would be used. In a real situation surrounding a ship in a bubble like energy construct would be impractical. To get a bubble only using one emitter, it would have to be at the ships center of gravity. that would mean projecting all that energy through the crew. Doesn't sound healthy. Placing multiple emitters on the hull and balancing them in a bubble shape, would waste energy. So shielding fields would extend from the ship in a pattern closer to that seen in infra red photography, or electromagnetic scans of an MRI.
What "looks" aerodynamic and what "is" are not the same. Underwater test of models in Japan prove this. 1701, 1701 (refit), 1701a, glide through the water with little effort or control. The Enterprise 1701-D still stable but requires effort to keep stable
USS Voyager requires massive modifications (clear plexi ) stablizers just to move in a straight line.
The farther away from the round saucer design, the less stable and aerodynamic the hull becomes.
BTW Other stable designs are Disney's Nauteliss, the Starblazer's Yamato, Seaview from Voyage to the bottom of the Sea.