MPEG4-SLS Streaming

[MRC01]

When it comes to high quality audio streaming, I've read a lot about MQA but not much about MPEG4-SLS. References, measurements, comparisons seem hard to find. The Wikipedia article doesn't say much. Primephonic's highest quality streaming uses the SLS format with Orastream, where Idagio's highest quality streams 44-16 FLAC. My first reaction to SLS is just cuz the sample rate is higher doesn't necessarily mean it's higher quality (one could apply lossy compression to a higher sample rate, just to light up the "High Res" icon in DACs for marketing reasons). Is SLS worthwhile is it just smoke & mirrors like MQA?

 

[amirm]

It is a lossy AAC codec layer with a lossless secondary layer. The latter is obviously is the same as the original. The lossy version is as good as AAC is which is very good.

It is not MQA in that it heavily inflates the file size to give you the lossless version. MQA does not do this.

At some very high level though, they are both layered encodings so some similarity exists.

 

[MRC01]

Ah, so it delivers the lossess layer if it can (if you have enough bandwidth). Otherwise if falls back to AAC lossy compression. Is that right?
If so, consider for example a 192-24 stereo source file. This requires about 9.2 Mbps or about half that if FLAC encoded. If you have that much bandwidth, then you get it. Otherwise, it falls back to AAC to use whatever bandwidth you have.

 

[amirm]

It depends on the scenario. For local playback, the file plays as AAC. Then a compliant decoder can extract the other layer and give lossless.

For streaming scenario you mention, it gets complicated. Is there a streaming solution for layered playback of SLS? If so, yes, it would work as you state. My sense though is that it is not a good fit for streaming. In that scenario server and client can negotiate and play native streams as needed. There is no reason to embed one stream in another when both are not delivered at once to the client.

 

[MRC01]

I've done some testing and SLS seems to work as advertised. My Ubuntu desktop using Pulseaudio has 2 configured bit rates, one primary, the other fall-back. It resamples audio when necessary to one of these rates. I set the rates to 88.2 and 96 respectively. This way if it does resample, it's always an integer multiple which simplifies the job and should eliminate side effects. As I play different tracks from Primephonic, I see Pulseaudio switching back and forth between 88.2 and 96 depending on the track.

Ideally I'd like it to use the native bit rate without any resampling, but this is the best I can do given the constraints of Pulseaudio.

 

[MRC01]

... Is there a streaming solution for layered playback of SLS? If so, yes, it would work as you state. My sense though is that it is not a good fit for streaming. In that scenario server and client can negotiate and play native streams as needed. There is no reason to embed one stream in another when both are not delivered at once to the client.

Yes, this is how Primephonic's audio streaming works. They use something called Orastream, which some other services also use.

BTW, I confirmed another thing you said above. The Primephonic app enables you to download albums locally so you can listen to them without any network. When you do this, it actually downloads the AAC compressed files. So the only way to get the lossless originals is over the network, if you have sufficient bandwidth.

 

[MRC01]

Here's a bit more on MPEG 4 SLS, in case anyone is interested: https://mpeg.chiariglione.org/standards/mpeg-4/audio
Scroll down to the SLS section.