I don’t think there are a lot of comparisons on the web (if any) concerning Dolby Surround decoding, so I felt pushed to do one! Without much further ado, let’s start!
First I needed a test clip; I found one with spoken channel names, so I used this [https://www2.iis.fraunhofer.de/AAC/multichannel.html] I discarded all except “front left”, “front right”, “front center”, “left surround, right surround” (I could include only one surround, I know, but if I included only “left surround” everyone would ask “where is the right surround?” and vice versa, so it is like it is!)
then it was fed into a simple Dolby Surround hardware encoder, that matrixed left, right, center and surround into LtRt, or Left Total Right Total; don’t forget that the surround channel is frequency limited to 100Hz-7000Hz, so the latter sounds different from the original
this is used to feed the following decoders; some are software, some are hardware; the former are all foobarr2000 plugins except the so-called “my way”, that is my personal interpretation of matrix encoding; the latter belong to a friend that kindly decided to do that for me; he simply sent me the files that I put in sync without telling me which brand and model he used, but in a way it’s even better, so not me nor you could be biased – at least about hardware!
I should remind that the matrix encoding and decoding process is lossy for its intrinsic nature and therefore it is IMPOSSIBLE to perfectly reconstruct the four (or more) channels once they are matrixed into two; that’s why, looking at the waveforms, often there are residual sounds on channels other than the proper ones – these are byproduct of the matrix process – and, unless very noticeable (when similar in level to the proper channels) their levels are very low and when played all at the same time impossible to hear.
This should be the “purest” decoded version of a Dolby Surround encoded track; it must sounded right nonetheless its limits and weaknesses – because the Dolby technicians listened carefully to any single track that was encoded via a decoder like this on the fly, and they should have corrected eventual problems whenever they occurred.
This was the successor of the original Pro Logic; it added stereo surround and better steering, and no frequency constrain on the surround; you can see cleaner waveforms, but not everywhere; this was not the one used to check encoded Dolby tracks and even if it should sound better, it would surely sound different and less “original”.
This is the first decoder not using the Dolby decoding method; you can see that its waveforms are even cleaner than the Dolby Pro Logic II decoder itself – probably due to improved electronics and/or different decoding algorithms.
Another non-Dolby decoder, different algorithm and electronics lead to different result; albeit cleaner on some parts, it’s worse on others – the left and right channels contains louder versions than other decoders on the surround parts. Notice that it’s the only one that uses the LFE channel.
This is a plugin that uses the original Dolby Pro Logic II library and should work similarly to the hardware version – that means well. Strangely, it is one of the worse! Take a look at the center – the second from the left; why it should be copied on the left and right?!? During listening this is easily detectable and quite different – in a bad way – from the others. I’m pretty sure it’s not a library fault, but just the plugin; more investigations are needed.
This is a mess… all channels are duplicated in others – frankly I can’t understand it! If you listen to it via any audio player, the sound that is supposed to come only from one speaker (two for surround) comes insted from three, four, or even five speakers at the time – what’s the point?!? And, if you play it on Audacity, the surround would be silent – of course, as left and right, both front and surround channels, are out of phase!!!
One of the most known free surround plugin, FreeSurround’s waveforms are really nice and clean, and it sounds good using any audio player; fact is, again, the surround are almost out of phase, so if you listen to them on Audacity, you will hear just a faint version of what you should hear…
This is my personal method of matrix decoding using only software in a very convoluted, difficult, long way, but result is worth it – it has the cleanest waveforms and most similar to the original (apart, of course, the frequency limitation of the surround channel); I’m proud of my work!
Hardware decoders, Dolby or not, ensure a good result; they are quite cheap nowadays, often much cheaper than software plugins, and if you use professional ones, that are built like a tank, you can get sometimes also digital output; but you need a multichannel sound card – or capture stereo in two or three sessions and then put everything in sync, not that handy – and two hours movie would take, well… two hours!
Software decoders: avoid the foobar’s upmix plugin, that is useless, and Dolby Pro Logic II plugin that sounds worse than the others; FreeSurround sounds good and it’s free, but its surround channels are wrongly out of phase, hence you simply have to discard one of them; my way of decoding gives the best result, but this is a secret recipe…
Of course this is a quick test made using a short audio clip with clearly discrete signals; a deeper test, using longer and more “difficult” source should lead to more detailed conclusions. Meanwhile, you can judge by yourself downloading all the clips you see here: https://mega.nz/folder/WsVCSLSI#FeFovcCOQDyb1Sr87JYZmA