Why does the sample rate not affect the number of CU‘s used?
We can say that in DAB+, the only two parameters that affect the number of CU’s are:
- The bitrate (going in steps of 8 kHz)
- The Forward Error Correction (EEP-A1, EEP-A2, EEP-A3, and EEP-A4)
The fact that the sample rate has no effect is related to the use of HE-AAC coding. As a general rule, the coder will automatically use the bits freed by a lower sample rate to increase the overall audio quality of the audio codec.
If a stream of 96 kbps at a sample rate of 48 kHz achieves a certain audio quality, switching to a sample rate of 32 kHz will free up bits. Because the sample rate only affects the highest audio frequency, the sound will miss some high-frequency audio spectrum between 16 kHz and 24 kHz. The coder will use the freed bits to enhance the quality of the 0-16 kHz band.
However, if the 96 kbps remain, the encoder will continue to use the same number of CU’s at a 32 kHz sample rate. The encoder will automatically use the additional bits to enhance the audio quality of the spectrum between 0 and 16 kHz. The question is whether the human ear will hear this increase in audio quality. Remember, 96 kbps at 48 kHz is considered the highest possible quality for HE-AAC.
Reducing the sample rate from 48 kHz to 32 kHz affects the highest frequencies but improves the quality of the remaining spectrum. However, since the quality was already acceptable, you could decrease the bitrate to gain more CUs.
Well, no rule in HE-AAC says how much you can decrease the bitrate. It must be determined by listening in practice to maintain the same audio quality. We estimate it would probably be around 80 kbps.
The conclusion is that reducing the sample rate from 48 kHz to 32 kHz will result in a reduced high-frequency component of the spectrum. However, you could also decrease the bitrate from 96 kbps to 80 kbps (or even 72 kbps) without any decrease in quality. Keeping the 96 kbps can be considered a waste of CU’s on the MUX capacity.
Assuming you have the above configuration for your station (80 kbps at 32 kHz). And you switch on SBR (Spectral Band Replication), you can reduce your audio bandwidth to about 8 kHz with a sample rate of 16 kHz, even if you still configure 32 kHz as the sample rate.
That’s what SBR does. However, SBR will synthetically increase the bandwidth back up to 16 kHz in the receiver’s decoder. So, you probably won’t hear any difference. Again, you freed up 50% of the bits for the encoder, which will increase the overall quality. It means you could decrease your bitrate again, perhaps to 64 kbps (or even 56 kbps) without noticing a difference in quality.
That’s how HE-AAC finally works. At 56 kbps, using separate stereo would sound poor. At such a low bitrate as 56 kbps, you will switch PS on (Parametric Stereo) to maintain the stereo level.
Additionally, switching PS on will prevent splitting the bitrate into two different channels; it will utilise almost 80% of the bandwidth for a single quality mono channel and provide an additional signal using the remaining 20% of the bits with the stereo information. The result is a better quality of the main mono channel. Again, with PS, you can achieve better stereo quality at lower bitrates.
The lesson learned is that by experimenting with sample rate, SBR, and PS, you need to understand that bits are freed in your stream, and as a result, you can gain CU’s by lowering the bitrate, while the quality will probably stay the same.
As said, there are no fixed rules on how much the bitrate can be reduced with a lower sample rate, SBR, and PS. It all depends on your appreciation of the sound. On Open Digital Radio’s github, we read the following info:
DAB+ AAC encoder configuration
By default, when not overridden by the –aaclc, –sbr or –ps options, the encoder is configured according to bitrate and number of channels.
If only one channel is used, SBR (Spectral-Band Replication, also called HE-AAC) is enabled up to 64kbps. AAC-LC is used for higher bitrates.
If two channels are used, PS (Parametric Stereo, also called HE-AAC v2) is enabled up to 48kbps. Between 56kbps and 80kbps, SBR is enabled. 88kbps and higher are using AAC-LC.
However, remember that only the bitrate and the EEP error correction will determine the used CU‘s for your DAB+ radio channel.