Speech must be good enough to enable the average native-speaker listener to correctly understand what they hear, by the speaker using the correct phonemes and intonation. Just good enough, no more.

Further details about this can be found here.