Does Gilbert Ryle’s distinction between knowledge-that and knowledge-how solve Mary’s room?
For Ryle there exists an often-neglected distinction between the two possible forms our knowledge may take. Firstly, there exists knowledge-that, of the form ‘know that’ which would be -->

followed by a declarative sentence (e.g. ‘I know the Earth is round’). This knowledge is semantically mediated and can essentially be transferred through mere communicative means i.e. it’s only necessary for me to tell you ‘the Earth is round’ for you to comprehend this fact -->

and possibly accept it.
There exists however a second form of knowledge, one that has arguably been neglected throughout philosophy due to some elitist biases or other factors, and that is knowledge-how, of the form ‘know how’ which would be followed by an infinite verb (e.g. -->

‘I know how to ride a bike’). The particular feature of this type of knowledge is that it possesses a necessary pragmatic component to it, it requires action, interaction, trial and error, for one to apprehend it, it cannot simply be transmitted by semantically mediated -->

communicative means. Although a conversation with an expert biker might give me some useful mental advice as to do’s and don’ts of riding a bike, it by itself could never be enough for me to learn how, or by that matter know how, to ride a bike. Knowledge of riding a bike is -->

non-reducible to semantics, it cannot be fully transmitted through mere communication, it requires the agent to actually participate in the action of riding a bike for them to truly acquire the knowledge of how to ride a bike, they could never do so simply by reading a book -->

simply by reading a book on how to bike or by that matter knowing the precise physical and anatomical interactions and mechanisms behind the movement of a bike and its user.
Now for Manuel DeLanda, who discusses this epistemic distinction in his lecture ‘Deleuze, Subjectivity -->

and Knowledge’, vision (as well as audition and other faculties) is actually a form of knowledge-how. The example he gives is the ability to differentiate between carmine and garnet, two distinct shades of red. Now if you don’t actually have an experienced eye, one that has -->

been trained to distinguish between the two, you most likely won’t be able to truly distinguish one from the other. Even if in the attempt to do so one person actually went and pointed at which red is carmine and which is garnet, and explained to you that the latter possesses -->

more purple in it than the former, you would just have the formal knowledge of which is which but your eyes would still be unable to truly visualize the distinction between both (this would also apply to recognition of some higher notes in certain music, an expert may explain -->

to you they’re there and enhance the experience, but you will likely not be able to hear them).
So taking vision to be a form of knowledge-how, wouldn’t this solve Mary’s room problem? (without the necessity of appealing to some non-physical qualia mind you)
Mary’s room asks -->

us to imagine a super-scientists who has never experienced the colour red but which knows all there is to know about its physical, chemical, whatever constitution, its relation to all other colours in the spectrum, and the biological reactions it provokes on the human eye -->

plus the brain processes that accompany such sighting. Now if she actually went and saw the colour red, actually went and experienced it, visualized it, would she learn something new, would she be under a new, previously unknown experience?
Well if we are talking about new -->

experiences, then even if Mary knew the compositions and interactions going on, the mere act of actually seeing red should obviously count as a new experience in everyone’s books, so it cannot be the case that Mary hasn’t experienced (i.e. hasn’t seen) red before through her -->

absolute physical understanding.
So the question would seem to rely on a distinction between seeing red and knowing what it is to see red. If there is such distinction however, the problem decomposes itself through its own premises as the actual sighting of red will count as -->

a distinct concept and thus a distinct and original experience. Therefore for the problem to withstand there must be no distinction between the knowledge of what it is to see red and seeing red.
The problem then would turn into: can Mary see red by having complete -->

descriptions (acquired through books, conversations, etc.) of all the physical, chemical, anatomical, etc. elements at hand? That is, can Mary see red (i.e. acquired knowledge of how to see red) solely through semantically mediated communicative means and no actual -->

interaction with the object of knowledge?
I think if we frame the problem like this, it’s very easy to substitute ‘can Mary see red by getting complete descriptions of how it is/what it is to see it through books, people, etc.’ with ‘can Mary see the difference between -->

carmine and garnet by getting complete descriptions of their differences/how the brain differentiates them from books, people, etc.’
The answer seems clear.