Which approach to speech recognition avoids the problem caused by the differences in the way words are pronounced according to context?

• continuous speech recognition
• connected word recognition
• isolated word recognition
• speaker-dependent recognition
• None of the above