Back when stereo audio for TV was developed, around 1984, RCA used, in addition to the regular video IF strip, a special video IF strip for the sound. One peak for the picture carrier, and another peak for the sound carrier. The "video" detector just beat these two carriers together to get the 4.5MHz sound IF, without incidental bandwidth tilt, which can cause that buzz, which will trash stereo decoding. There's no Nyquist slope at the picture carrier, like regular video IFs have.
That Sony with the choice of sound IFs sounds
quite interesting. I suppose the tuner was a PLL based on a crystal to yield an accurate local oscillator frequency, so a split sound system would always work. They could have used the FM detector to provide some automatic frequency control correction (probably to induce a slight shift in frequency of the crystal reference oscillator, to shift the local oscillator as a result of the PLL).
Oh, I see VK plants web page links and $ coins on words like "Sony". Well, if the money Sony and others pays for it helps keep VK on the air...