|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
I was just reading a paragraph in a review of the Alesis, and the guy said the following:"In a traditional workstation, a track is scanned for its peaks...'digital zero' is determined, and, if desired, the level of the entire track can be raised by that amount and, typically, a new file is rendered."
I was thinking that when recording to the Masterlink, one would have to figure out where the highest-level peak signals were BEFOREHAND and use that level as your 0 dB reference. In this way, the recording would maximize the number of bits used to record the track in question. I don't see how you can record to the HD at some "safe" lower level in the first stage, and then go in and have the Masterlink boost everything by N decibels. Does this practice equate to sacrificing some resolution?
[BTW, In my application, I'm talking about an analog source that you would be converting to digital on the fly, as you're recording to the Masterlink, so there are no digital files that can be 'scanned' for peaks beforehand.]
OR.....do they assume you're not a complete moron and that you're leaving a safe but not HUGE amount of space under the 0dB ceiling on the first pass, and since you'd have to be under it by *more* than 6dB before you would subsequently end up "losing" that first bit of resolution, you don't really have to worry about it? (--for two reasons: one, anyone with a brain can play a few seconds of the highest-level passages and have a feel for how close they should set the recording level to be in the 6dB ballpark; and two, losing 1 bit when you're working with 24 bits is trivial anyway, because there's no playback system that can resolve a full 24-bits of dyn range anyway)
Any thoughts/wisdom on this?
:^)
Thanks in advance for any info provided--
Mike
Follow Ups:
Normalization doesn't magically restore bits that were lost in the recording process (for the sake of headroom), no. It's just digital hygiene.Keep in mind that '24-bit' converters are not really 24-bits, so when you lose a bit, you're not losing the least significant bit (you can't lose what you never had in the first place).
re: 24 bits not really achievable... gotcha -Before you start losing potential bits of resolution, how much headroom would you have to leave? If you record so that after going back and inspecting the digital file there is only 4 or 5 dB of headroom between the highest peak level on the track and the 0dB max level, have you lost potential resolution? I am wondering if you need to be more than 6 dB under before you lose one bit of rez.
?
Mike
If you record so that after going back and inspecting the digital file there is only 4 or 5 dB of headroom between the highest peak level on the track and the 0dB max level, have you lost potential resolution?Yes, 4 or 5 dB. Some loss is unavoidable, and (of course) is far preferable to clipping distortion.
Is there any chance that you can check the tapes first to see what the peak level is? The more bits you record to start with the higher the resolution of your recording. Well recorded tapes should come with reference levels shown on the box!Regards,
AndyG
Good point, but I'm posing this as more of a hypothetical, for when you have an analog source that's either not referenced, or else you're doing a transfer from LP, etc.If you do a little previewing of the material and are skillful enough to leave *less* than 6dB of "room" between the highest peak level of your material and the 0dB [digital maximum] level, would it be correct to say that you're probably losing no more than 1 bit? (Which at 24 bits is not a huge deal) ?
Mike
If you are copying from a record, your dynamic range is about 56db at the very best. A 16 bit recording has a theoretical 96db,(typically 92 in real terms). 24 bit recordings will well exceed your analogue input electronics, the best SN ratio I've seen quoted on 24 bit converters is 107db. If you run a little test you should be able to align your peak within 2db of digital 0db without any problem.Regards
Roland
Is there something other than taking a few minutes to play what appear to be the widest groove modulations on the record, and see where they peak out? If it's not too involved, what is the test you're referring to..? Thanks for your help -Mike
Thats exactly what I'm referring too! You wouldn't go to lay bricks without mixing up mortar, why on earth would you try to record without taking a level. On live classical sessions I regulary mod up to within about 1-2db peak level on a live session without much dificulty. As I said before, coming off records your dynamic range is horrendous, so normalization or not isn't going to have any audible effect except that of level. But that is no excuse for poor recording practice.Regards
Roland
This post is made possible by the generous support of people like you and our sponsors: