+1 to a batch process that you can adjust threshold for. Then you can simply review tracks for content and ease the process.
As an aside, you mentioned track length, implying cost is based only on time and not word count? If so, another thing you can do is bounce the track so it is one clip then stretch-edit to shorten it (ctrl-drag the right edge, and bounce again). In general, people can understand 3 times as quickly as they speak, but don't go overboard with such if you choose this (can probably save 10-20% right there).