Another thread on digital compression

[Werner]

I've just read the thread refered to at the start and something AS posted makes me think he either has no idea how digital recording works or I don't.

I think AS got a bit carried away in his argument, when he stated that for bass 6-8 bit would be enough. But I also think he was indirectly referring to the ear's lack of sensitivity there. When sliding down from 100Hz to 10Hz the threshold of hearing increases from 25dB SPL to 75dB, which could be expressed is an equivalent loss of resolution of 8 bits, give or take.

 

[Coda II]

I think AS got a bit carried away in his argument, when he stated that for bass 6-8 bit would be enough. But I also think he was indirectly referring to the ear's lack of sensitivity there. When sliding down from 100Hz to 10Hz the threshold of hearing increases from 25dB SPL to 75dB, which could be expressed is an equivalent loss of resolution of 8 bits, give or take.

Given that you seem to have a fair understanding of where AS is coming from, would you be able to give a brief summary of his views in that thread?

Not putting words in someone else's mouth (and obviously not getting into areas that would be against the AUP here) but just the basic stuff like are compressed formats a 'good thing' or a 'bad thing'?

Via that thread I ended up with a pretty fair understanding of the concepts behind the mp3 codec as well as masking and psychoacoustics, but apparently it went off the rails.

 

[Labarum]

but apparently it went off the rails.

I am glad you said apparently.

My point?

Listen to John Stanley's Trumpet tune.

There is an MP3 here. Choose the Audio Tab and look down the page

http://www.makinorgans.co.uk/Downloads.aspx

This is a recording of a digital electronic organ that plays processed samples of real organ pipes. The organ loudspeakers would have been used in the cathedral and the sound recorded by microphones.

The trumpet stop (a rank of pipes producing their sounds by vibrating metal reeds) is played from one manual (keyboard) and the accompaniment is played on another manual.

I suspect the trumpet stop does have high harmonics; and I guess this simple organ piece is a good subject to illustrate masking.

The much more demanding organ work below must be an MP3 killer, indeed 'twill devastate all but the most highly specified audio system.

http://wn.com/Louis_Vierne__Carillon_de_Westminster

This piece I have heard a number of times played on the York Minster organ which has probably, the loudest tuba in the world. This rank of reeds runs at nearly four times the air pressure used by the routine stops on the organ. (15 inches measured on a water manometer)

Don't tell me those reeds aren't generating significant energy up into the supersonics.

Of course NO hifi can do such a sound justice - you have to be there.

 

[Labarum]

Another question on masking I wanted to ask concerned the Harpsichord, and even Haprsichord continuo (background playing.)

This sort of thing

The Harpsichord is not a powerful instrument. A modern Concert Grand Piano can hold it's own with an orchestra, but the harpsichord makes a tiny sound and yet remains audible.

In the illustration above the harpsichord is in the foreground quite a lot of the time, it is a Harpsichord Concerto, but it illustrates how the tones of the plucked strings cut through the other stringed instruments. In other pieces it just tinkles along in the background - continuo playing - with, maybe, even a choir and organ to fight off. Even when the going gets quite tough the harpsichord continuo remains audible.

I wonder how MP3 masking handles this?