It depends on how it was converted. Many OCR applications somehow place text under the image. Some do this by first putting the text down, placing the image on top. Some place a snapshot at the bottom, and then put the text on top, using the “Do Not Mark” transmission mode.
I mention this because I cannot predict how any text selection tool will respond to transparent text. Theoretically, it should just give you text (this is what Acrobat does). Does anyone realize if this is really happening in all text extraction tools.