Audio bit rate

Originally Posted by MarkHotchkiss 

Hi Kevin,
As you and Mcnaris stated, 24-bit isn't necessary for the finished product, but might be necessary for mixing. If you have something recorded at 16-bits, and then turn it down 12db for the mix, then you had better have a few bits under those top 16 to catch the two bits you would otherwise loose. 16-bits in and 16-bits out is adequate, but more may be needed for the processing in between, to handle math overflow and underflow. That's why audio DSPs have at least 24-bits.

Originally Posted by mcnarus 

Uh, no. CDs are 16 bit. Period. There are some recordings out there that are available in 24 bit. But they aren't on CDs.
It isn't, in part for the reasons you give. But there's this whole audiophile mythology that grew up around the idea that there's something wrong with CDs. So anything that comes along that claims to be better automatically has credibility among the myth-believers.

Originally Posted by amirm 

You are measuring it wrong . You have to use psychoacoustically correct measurements or you get the wrong dynamic range. I explain it all in my article on dynamic range here: http://www.madronadigital.com/Library/RoomDynamicRange.html

Net summary, 120 db is the channel we need for transparency for all content and all rooms.

Originally Posted by arnyk 

Actually, an overly enthusiastic article that defies common sense, was amended by later science, and failed to produce salable products when people tried to implement its ideas. It represents an unreachable goal.

First off, a 120 dB dynamic range is unsafe for humans if ever implemented. In an ideal world, the threshold of hearing is around 0 dB SPL, while any amount of listening at 120 dB will either permanently damage or temporarily deafen your ears. In the real world, rooms with acoustic noise levels of 25 dB SPL are difficult or impossible to implement in ordinary residential circumstances, Most rooms in a house come closer to 35 to 45 dB SPL range. But lets momentarily accept 25 dB SPL. 120 dB dynamic range implies that the louder passages play at 145 dB SPL. SPLs on this order are harmful or fatal when unprotected humans are exposed to them without ear protection and protective clothing.

Secondly, there is a widely-recognized international standard for audio evaluation known as ITU recommendation BS 1116-1 which represents later science and art than those few earlier papers favor. It basically mandates 100 dB dynamic range which is of course far less demanding and more realistic than 120 dB.

Thirdly we had three formats for recorded music that were designed to meet or approach the 120 dB standard being HDCD, DVD-A and SACD. All three were invested in heavily by large audio firms, large number of players embodying their technology were sold, many no doubt at a loss to their manufacturers. Large numbers of recorded releases with their logos on them were put into the marketplace. None were successful in the mainstream, and all three are in various stages of disappearing.

Finally, there just aren't any commercial musical recordings that actually have 120 dB dynamic range. Their media may support it but quite a bit gets lost when people actually try to record music. The widest dynamic range musical recording I've been able to find in about 20 years of searching has less than 90 dB dynamic range. If you set up some microphones in a typical concert hall in accordance with the best accepted standards, bring in about 100 musicians and record something they play, you end up with a recording with something like 65-70 dB dynamic range.

Originally Posted by Ethan Winer 

Just to clarify, modern audio software doesn't work that way. 16-bit files (or any files) are loaded into RAM, then converted to 32-bit floating point for all subsequent manipulation. You can have a 16-bit file, reduce the volume by 100 dB, apply EQ or whatever, then raise it back up by 100 dB and it will still sound fine. The quote below is from my Audio Expert book.

--Ethan

Originally Posted by amirm 

That would be the wrong measurement too . As with the noise floor of the room, you need to look at the spectrum of the recorded content's noise floor, and then apply the math to covert the noise to equiv tone value. Once there, you can then compute what the dynamic range is for the frequencies where the ear is most sensitive. Your SPL meter or whatever you used doesn't do any of this analysis and hence produces vastly incorrect values. This is why we call them "dumb meters."

While typing this, let me remark on the ITU BS.1116 comment. The ITU recommendations is for setting a set of standards for testing audio impairments. It is not a document that sets out the standard for reproduction fidelity. Here is what it says about dynamic range anyway:

"7.2.2.6 Dynamic range
The maximum operating sound level which the loudspeaker can produce for a time period of at least 10 min without thermal or mechanical damage and without overload circuits being activated, measured with a programme simulating noise signal (according to International Electrotechnical Commission (IEC) Publication 268-1c), should be:

L eff max > 108 dB"

So not sure where the 100 db reference came from and at any rate, 100 is > CD anyway.

As for going deaf, that is a function of both level and duration. And at any rate, if I am worried about going deaf, I can turn the volume down. I don't need someone to dumb down the dynamic range on my behalf .