Topic: Analog to 24bit question

[datbrad]

I have been reading this thread and all the points being made have really made me think, which is really the point with honest discourse anyway. I backed away from the deep detail of the finer points of sampling theory and word length qualities and came up with what I think is good answer to the original posters real overall question.

A 16 bit recording with noise shaped dither sounds basically no different to a human ear than a straight 24 bit quantitized recording with the same gain level unless the dynamic range of the source exceeds about 114db. A 24 bit signal converted to 16 bit using a noise shaped dither such as UV22 will produce an effective dynamic range of 19 bits, since the quantitization noise is taken out of the 24 bit data before it is converted to 16 bit. A straight 16 bit master leaves the quantitization noise intact and is more audible than the noise left intact on a straight 24 bit recording. This is because the quantitization noise for 24 bit is below the analog noise floor of real world equipment, but not always so for straight 16 bit.

Here is where we get to the meat of the original poster's question. It's not about the final playback of the converted cassettes, it's about the recording of them to digital. Unless he is recording the cassettes using an A/D with real time noise shaped dither, like an Apogee, to master at 16 bit, he must use 24 bit to master and then apply the filter in the post conversion, using UV22 in Wavelab.

Then, it would up to him to retain and store the original 24 bit master file, or just the final noise shape dithered 16 bit file, since they would both sound basically identical with a cassette as the source.

Hope this helps!

[easy jim]

For me and for all my practical purposes I will stick with 24/48; maybe a 24/88.2 sometime, but I doubt it. 

I have come to the same conclusion as far as 24 bit recording goes...that is unless I switch to DSD if/once it catches on more than now.

[F.O.Bean]

Sorry Petrus, not directed at you.  I seem to live in a disabled world on this board where reply to doesn't work and quoting has to be done manually, so I skip it.

I can state what I can hear as well as what I can measure (D/A conversion is not required, just some analysis with a good SRC.  But a D/A/D analysis is not troublesome for a reasonable converter either).  I can hear the difference between 44.1 and 48; I expect that anyone with enough remaining high-frequency hearing* who knows what to listen for could hear it.  But it's rather subtle.

I can't hear the difference between 48 and any higher rate, but I can measure a difference in the audio spectrum between 48 and 64, but not any higher rate than that.  So Lavry's theory (also supported by other EEs who design converters; see Bruno Putzeys' board on PSW for examples) can be quickly shown to be measurable even without resorting to listening tests.

But the statement that higher rates (both sample and bit depth) can always only be good is false and unhelpful at a time when the thread had seemed to come near a conclusion.




* The other day for fun I tested my daughters' (7 and 10) hearing at 19.5kHz.  I'm sorry I didn't measure the volume accurately, but it would have been between 70 and 80dBSPL.  There were no lower-frequency distortion products anywhere near that SPL level.  I could not detect it at all (I frequently test my ears, and I'm good to 17kHz in one ear, and 12kHz in the other at 70dBSPL--ear infection damage and tinnitus in the one ear).  Both girls passed the test easily, without even hesitating when the tone started.

Perhaps I should give them the 48 vs 96 sample rate test . . . I can test up to about 22kHz accurately through my playback system . . .

sorry for the highjack, but you said you cant quote/reply? Do you have Firefox installed? If so, disable the "No-Script" for ts.com and they should both work fine....

[Keyd]

Interesting info in this thread.

Thanks folks.

[heath]

You are probably not going to find any studios who get paid for transferring DAT masters -- what bands used 2 track DAT for studio work? -- maybe ADATs but..., thus very little online discussion about the best methods of editing ADATs.

I do     

sorry to jump in late on this one.  seems like it got rather heated...  an interesting read nonetheless. 

[DSatz]

DATBRAD, the thing is, for a transfer from cassette, 16-bit with garden-variety neutral dither gives you way more than enough dynamic range. There's no need for UV 22, or 24-bit, or whatever. I mean for God's sake, 16-bit linear PCM with garden-variety neutral dither gives you more than enough dynamic range to handle 15 ips half-track Dolby "A" open reel master tapes already.

What is the problem here? Why do people imagine that you'd need to strap yourself into a frikkin' Atlas booster rocket just to cross the street in your own little neighborhood? It's a cassette, people. Take any that you have, with the widest dynamic range recorded on it, transfer it to 16-bit so that the peaks are at a nice, comfortable -3 dBFS, and the noise floor of the cassette will be 30+ dB above the noise floor of your transfer.

Just to make sure, I just now did what I described, and I'm seeing levels of around -60 dBFS on the blank part of the tape with Dolby "B" on and 70 microsecond playback EQ selected. I mean, 14 bits would be more than enough for that; 12 bits would be enough.

I do not, do not, do not get this whole "princess and the pea" attitude. (goes offstage muttering to self ...)

--best regards

[George2]

I hope that puts an end to this drivel. Well said.