There's a huge difference in the sentences, but how you would speak either out would be exactly the same. When we speak to each other, we don't "see" apostrophes either. Our only hope in the case above is context, and as Shyflower said, context can be ambiguous.I don't really think your example shows the difference.
When using speech-to-text for creating code or composing messages, there needs to be a separate mode or set of commands for setting things like punctuation and grammar. So after stating "there was no food in the house, so I ate the dogs" you would need to go into an edit mode to manually add the apostrophe before returning to writing mode.
Any speech program that doesn't have something like this built in is a fail.