This is defined in ISO/IEC 13818-3 as:
2.1.10 audio access unit [audio]: For Layers I and II, an audio access unit is defined as the smallest part of the encoded bit stream which can be decoded by itself, where decoded means "fully reconstructed sound". For Layer III, an audio access unit is part of the bit stream that is decodable with the use of previously acquired main information
How can I determine the duration of audio access units in any given MP3 file?