Passage Summary: Computers can't tell 'their' from 'there' yet, so they won't be able to write down what we say correctly until they learn how sentences and word meanings work.

Conclusion: Voice-recognition programs will fail to accurately transcribe speech until they can understand grammar and the meaning of words.

Reasoning: Current technology is unable to tell the difference between homophones, which are words that sound identical but have different meanings and spellings.

Analysis: The argument identifies a specific technical hurdle—homophones—and concludes that a very broad goal—accuracy—cannot be met without a specific solution—semantic and grammatical understanding. The 'Gap' here is the assumption that you cannot have an accurate program if it fails to distinguish between words like 'their' and 'there.' To find the necessary assumption, ask yourself: what if a program could be 99% accurate even if it occasionally missed a homophone? The argument relies on the idea that distinguishing these specific words is essential to the definition of 'accurate' transcription.