(We would like to remind the reader that DAB/DAB+ uses a sample rate of either 48 kHz, 32 kHz, 24 kHz, or 16 kHz).

Superframe packing was a necessary enhancement to adapt the original DAB system for HE-AAC audio coding in DAB+.

The original DAB standard was designed around audio frames of 24 milliseconds. However, this fixed duration was not compatible with the frame structure of HE-AAC, which required a different frame length. As a result, a new solution was introduced to reconcile these differences.

For example, we use the most common sample rate in DAB+, which is 48 kHz.

When transforming audio from the time domain to the frequency domain, it is common practice to use a number of samples that is a power of two. A frame size of 1024 samples would be a logical choice. At a 48 kHz sampling rate, this corresponds to a frame duration of:

1024 / 48,000 = 21.333… ms

To achieve a cleaner match, we consider reducing the frame size to 960 samples, which results in a duration of:

960 / 48,000 = 20 ms

However, even with this adjustment, we still face a mismatch with the DAB system’s native 24 ms frame duration. The key insight is that the least common multiple of 20 ms and 24 ms is 120 ms.

This allows for an elegant solution:

• 5 frames of 24 ms = 120 ms
• 6 frames of 20 ms (HE-AAC) = 120 ms

By grouping six 20 ms HE-AAC frames, we form a superframe of 120 ms, which can be distributed across five 24 ms slots in the original DAB frame structure. This process is known as Audio Superframe Packing, and it is a DAB+-specific adaptation.

Conclusion:

The Audio Superframe Packing process takes place after HE-AAC (v1/v2) audio encoding. The Reed–Solomon coding (RS(120,110)) is then applied to the packed superframe to provide error correction. This entire process is part of the application layer in the DAB+ architecture, as reflected in the relevant system diagrams.

What about the other DAB+ sampling rates?

We have observed that at a 48 kHz sample rate, we utilise six frames of 20 ms each to fill the superframe.

If we examine the other sampling rates of DAB+, a similar packaging approach can be applied.

• 960 samples at 32 kHz result in frames of 960/32 = 30 ms. Using 4 of these frames, we again have a superframe of 120 ms.
• 960 samples at 24 kHz result in frames of 960/24 = 40 ms. Using 3 of these frames, we again have a superframe of 120 ms.
• 960 samples at 16 kHz result in frames of 960/16 = 20 ms. Using 2 of these frames, we again have a superframe of 120 ms.

Back