Image Description Guidelines from DIAGRAM
DIAGRAM (or Digital Image And Graphic Resources for Accessible Materials) is a research center that produced one of the most useful image description guidelines for both the fundamentals of image description and the more advanced, complex image description that you encounter in higher education. I used these for describing everything from classic art pieces to chemistry flowcharts.
In some ways, you are participating in some high level Design Thinking when you are writing image descriptions. You are not just thinking of the physical characteristics of what you are looking at - you are thinking about the authorial intent, the intended audience, the function, the semantic impact of the image, and more. It can be pretty intimidating! Which is why we have these sorts of handbooks. It takes skill and practice, though if you are the actual author, you can avoid having to reverse engineer the intent, which is often the hardest part of image description.
MathSpeak™ Core Specification Grammar Rules
SeeWriteHear stewards what I knew as gh's MathSpeak. This is a set of grammar rules for verbally describing math which is based on the Nemeth Code used for braille math. Interestingly, this is what taught me about how poorly constructed visual math is- translating it into a parseable set of linearly read grammar rules is a nightmare, due to the adhoc way symbols are incorporated or repurposed to mean entirely different things depending on the context of the math.
I had to learn MathSpeak so I could write the alt text (not literally held in alt attributes, but actually visually hidden text in HTML documents) for math equations that were rendered using SVGs - entirely a backwards way of making robust digital math in some ways, but necessary when students needed access to their homework and couldn't wait for browser vendors to support MathML or for screen reader software developers to do the browser vendor's work for them.
Learning MathSpeak is useful for double-checking the way you have written your MathML or how your MathML is being rendered, because what looks visually correct can often have semantically drastically different meanings, which is extremely dicey when it comes to math homework!