In the other thread, I did another test (linked) with the mentioned A827 preset and 24 tracks (larger machines are not known to me). It turned out that on a k-weighted meter, the summing bus can reach -68dB RMS (-71dB RMS unweighted), if the noise is pulled back to -100. Each individual channel then has -82dB RMS. Which in old days, was unrealistic.sascha wrote:- we can make the hiss control go down to -120 or lower (although I find it ridicolous and counter-productive, but anyway)
Which still gives me the impression, that the channels were only measured in mono, stereo or 4 tracks at max to let the machine look better on paper.
With individual channels being at -68dB RMS (k-weighted), the summing bus would result in -53dB RMS (k-weighted). Still suitable and as expected with tapes.
A noise gate would be suitable IMO. PLenty of fish in the sea.sascha wrote:- we will think about auto-mute for the hiss control (doesn't mean we do it, at this stage)
As I said, this might not be possible at present, and lord forbid that you are the first dive into this very delicate topic just to please everyone. (pretty much everyone caved at this)sascha wrote: - we won't make hiss global. First, we can't. There's no way to collect several parallel buffers of their instances, at least not with the APIs we know. Otherwise we'd easily implemented crosstalk with neighbouring tracks (not only stereo per instance). Second, we wouldn't want to. This is not the way physics works. Hiss is random orientation of the particles and alteration of the remanence field due to their uneven structure. This greatly interacts with the source signal, especially since we're using a model consisting of virtual coils, hysteris loop, HF bias and eddy currents / self-erasure. Putting static over a global signal does not interact like individual noise and is pretty much like a song with vocals compared to karaoke (to draw an analogy).
There are ways around it. You just need to know them, or finally realise that a tape machine is simply noisy. And that you shouldn't be surprised that if you link 32+ machines together (to simulate a 32track environment) is bound to produce a lot of noise!
Which can be covered with a plain noise gate on the summing bus!
