The following tips are taken from Humber College's Captioning Style Guide.

Spelling

• Spelling should always be done to Canadian standards. 

Punctuation

• Serial or "Oxford" commas should be employed for clarity. A sentence like, “I like peanut butter and jam, and tuna sandwiches,” conveys a much different meaning than: “I like peanut butter and jam and tuna sandwiches.”
• Parentheses should not be used when transcribing text because they are used to denote content that is not narration.
• Dashes (-) should be used to end a sentence that is cut off or interrupted. If a speaker interrupts themselves or trails off, use ellipsis (...) instead.

Filler

• Delete filler like "um," "uh," "er," and other sounds that are generally meaningless.

False Starts

• Words and phrases like "so," "well," "you know," and others can be removed if it will not change the meaning of the text.
  “So, you know, if we take a look at this example...” → “If we take a look at this example...”
  “When, when you consider...” → “When you consider...”

Lectures & Tutorials

When an instructor is describing what is happening on screen, it is important to make the transcripts consistent and easy to understand.

• Alphanumeric keys are presented as capital letters, numbers, or symbols (A,X,1,0,?,/).
• Cursor and navigation keys are named and capitalized (Left Arrow, Page Up, Backspace, Insert, Home, End, etc.)
• The modifier keys are Ctrl, Al, Shift, Caps Lock and Windows (or Win) Key, Command and Option.
• Key combinations are presented with a "+" between the modifier and the key being pressed: Ctrl+S, Command+W. Certain symbols are spelled out in this context, such as Command+Period.
• Capitalize and spell menu selections as they appear on screen, separated by commas: To save your document, click on File, Save As.
• On-screen elements should be written as they are on the screen but only when they are referred to specifically: To create a black-and-white photograph, go to Image, Adjust, Black & White.
• Canadian spelling is to be used throughout a tutorial except when the speaker is specifically referring to an on-screen element – in Photoshop, you pick a colour using the Color Picker.
• “Okay” is used when transcribing speech, but click OK for the on-screen action as that is how OK is displayed in most computer programs.

Slang & Contractions

• Change slang like "gonna" and "wanna" to "going to" and "want to."
• Contractions should be transcribed as is whenever possible: “wouldn’t,” “isn’t,” “could’ve,” even “ain’t.”

Numbers

• Numbers from one to nine are written as words unless they are part of an equation.
• Words are also used for uncertain or estimated amounts: Thousands of people paid hundreds of dollars to see dozens of animals.
• Use "per cent" rather than the % symbol.

Profanity

• Any profanity or “swear words” should be captured as they are spoken.

Bad Quality or Missing Audio

• Use (indiscernible) when the audio is there but you can't make it out.
• Use (inaudible) when it is completely missing from the recording.
• Use (no narration) when there is an extended period of silence.

Identifying Speakers

• If the names of the speakers are known, put the name of who is speaking in the caption in capital letters followed by a colon. Use first and last name for the speaker's first appearance, only the last name subsequently.
  BOB SMITH: I'd like to introduce Jane Jones. JANE JONES: Hi everyone. SMITH: It's great to have you.
• If speakers' names are unknown, choose suitable descriptors for each speaker: SPEAKER 1, SPEAKER 2, etc.