Quantcast

SQ Decoding for binaural rendering.

Help Support QuadraphonicQuad:

Flingle

New member
Joined
Aug 5, 2020
Messages
2
Location
UK
Hello. I’ve spent a some time lurking in the background of this group, this is my first post, sorry its a long one.
I don’t have a Quad set up at home, mine is in the office. I’m an audio engineer for a large broadcasting corporation, in their training department, more on the technical side than the operational side, though I have spent many hours in studios, mainly lying under or leaning over the innards of various mixing desks or tape machines. I have been fascinated with binaural since I was a child after reading an article about it and Holophonics in a magazine (Electronics Today International, ETI). I now work with some very interesting and cutting edge Binaural steering software and teach our operators and engineers about it. I have built and recorded with many dummy heads. My question is about SQ coding for a purely personal project. I’m a fan of a 60s band who after they initially split in 69 some members formed a vocal harmony acoustic group (Marvin Welch and Farrar), they released 3 albums one of which was Quad, recorded for EMI so it was in SQ. I have got a good copy of the album and want to listen to it at home. I have used a software decoder called SQ Decode by Marcus Junglas. It produces a 4 channel WAV file from an SQ WAV file. I played the album into the software and out dropped four channels, my plan was to then use Sennheiser’s Ambeo Orbit software to make a binaural version to listen to on headphones at home. However there may be a problem with cross talk or odd decoding. Has anyone used SQ Decode and found problems with it? I did some tests and found that Rb is missing large parts (frequency and amplitude variations) of the expected input on that channel and has lots of cross talk from the other channels. Is this a trait of SQ? When I listen to the album on stereo speakers the treble sounds very “brittle” as if on the edge of phasing.
I made a test file with 12 tones, 3 for each channel, they play the A minor 7th chord in three octaves ( a minor 7th chord has four notes not three, so the tones should not drift into and out of coincidence as they are oddly related). I then use Audacity to create the Lt Rt using the formulae Lt= Lf -0.71j Lb + 0.7Rb, Rt = Rf+0.7lb + 0.7jRb . I use sine waves so the phase shifting is simply a matter of cutting part cycles at the beginning of the tracks. Rb’s output should contain three tones at normal level and some expected cross talk. However, the expected tones are 6dB down and the other tones (which shouldn’t be there but I know to expect quite a bit of cross talk from SQ) are there 3-5dB higher than the wanted signal. Any thoughts? Further to this I have made another test file, this time using square waves one frequency on each channel (again Am7) then after 30 seconds of tone there is 30 seconds of pink noise, to identify which channel the noise is from I put 18dB notches at different frequencies. After forming LtRt and passing through SQ Decode I get similar to before .Cheers.
 

Flingle

New member
Joined
Aug 5, 2020
Messages
2
Location
UK
Hi Ooops. Thats a typo in the post. I do use the formulae you quote. Thanks
 
Top