key, mode and confidence have been part of the audio_summary for years. This tutorial provides more info about the audio_summary attributes.
In your call, you are setting the bucket parameter twice, to two different values, which is probably not what you want.