|
Post by Martin John Butler on Jan 12, 2016 10:05:05 GMT -6
Tony said, "and make sure you dither on bounce.". Uh oh, I have to plead guilty on all counts, because I've turned off dither when bouncing because I don't completely understand the process, so I avoided it. Could that be one of the causes of my edgy stressed out mixes?
|
|
|
Post by jimwilliams on Jan 12, 2016 10:56:31 GMT -6
Peak levels out of my console hit +14 dbu, about where I want it. That way I have 6 db headroom in the PCM4222 ADC and the signal is several db further above the console's residual noise floor (already very low). THD specs also improve at higher levels for the analog electronics.
I run it all very hot here because it measures better and sounds better. Nothing is ever clipped here, that's saved for my fuzz boxes and guitar amps.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 12, 2016 11:13:49 GMT -6
MJB: Is your music hitting a DAC? Dither. Is your music written to a 24bit or 16bit or any integer file format? Dither. That's all. And yes, it *can* sound a bit edgy without. It becomes worse, if there is another e.g. any digital gain stage afterwards, because the quantization error becomes more obvious often. Use triangular dither. It can be used in all places, including final dither. Also the mandatory dither for everything that is to be processed digitally afterwards. So i followed Bob O.'s advice and voila - i think it sounds better than the noiseshaped ones.
|
|
|
Post by Martin John Butler on Jan 12, 2016 12:38:15 GMT -6
Whew, tanks SBF. I use Logic X, do you know if there's a choice of dither type? I don't recall seeing it. I've gotta run to work now, or I'd pul up a track and look for myself.
|
|
|
Post by wiz on Jan 12, 2016 15:10:44 GMT -6
Peak levels out of my console hit +14 dbu, about where I want it. That way I have 6 db headroom in the PCM4222 ADC and the signal is several db further above the console's residual noise floor (already very low). THD specs also improve at higher levels for the analog electronics. I run it all very hot here because it measures better and sounds better. Nothing is ever clipped here, that's saved for my fuzz boxes and guitar amps. The point in Jims post here... is that he has 6dB headroom.....and the console is operating where it sounds great as well as performing technically great. cheers Wiz
|
|
|
Post by mrholmes on Jan 12, 2016 17:09:09 GMT -6
Whew, tanks SBF. I use Logic X, do you know if there's a choice of dither type? I don't recall seeing it. I've gotta run to work now, or I'd pul up a track and look for myself. You dither before going down to a lower bit depth. If you press bounce a pop up float window is displayed, choose WAV 16 BIT 44.1 khz and one of the dither options.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 12, 2016 18:28:11 GMT -6
That's actually quite not completely true (or maybe i understood you wrong?)
I tried to explain to the Mixbus guys at Harrison, too, because the dither is greyed out at 24bit bitdepth export in Mixbus, which i consider is a technical error, which is actually not good at all. While the audio is in the DAW, and only the slightest bit of processing is done, a dither will be needed at export to everything but the native format of the DAW. Which is nowadays 32 or 64 bit FLOAT, sometimes it can even be changed from one to the other like in Sonar. So if you write the file from the DAW to something like 24bit, or to a DAC, which also ends in an integer format, you NEED dither, because there is requantization going on. The format is always higher resolution in the DAWs float formats for processing. Even if you started with 16bit or even 12 or 8 bit recording files. Only exception would be, that you did not process the input file AT ALL. No gain, no format change, nothing, and going out with the same bit depth. In this case the use of a DAW is completely pointless - in and out identical. I guess there is so much confusion from the time when the internal DAW formats were integer, like the input files and the targets. There, things were a bit different. But well, we have 2016 now...
|
|
|
Post by M57 on Jan 12, 2016 19:42:07 GMT -6
In Logic options at 24 Bit are: POW-r #1 (Dithering) POW-r #2 (Noise Shaping) POW-r #3 (Noise Shaping)
if you lower the resolution to 16 Bit you can add: UV22HR
So from what I'm reading here and elsewhere, basically it sounds like POW-r #1 is gonna be your standard choice - assuming you're keeping things at 24 Bit. But then say you wanted to post to SoundCloud - Are you perhaps better off bouncing to 16 Bit (using Logic's converters?) .. and then you have to choose between POW-r #1 and UV22HR. My guess is you're probably better off uploading it at 24 Bits because SC probably uses a proprietary conversion algorithm anyway you slice it.
|
|
|
Post by mrholmes on Jan 12, 2016 21:48:18 GMT -6
Thanks mrholmes. The thing is, those specs didn't matter to me until I began to understand what's needed from all the helpful posts here. I basically understood to stay under -12 on my DAW's meter, which I of course ignored all too often. I'll try to be smarter about the process from now on, and ask for help here when I need it. I know the cats here won't let me down, or laugh at my questions, though I may cause a chuckle or two now and then. Correct me if I am wrong but given that you set - 20 dbfs equal to + 4 dbu it means you have 20 dbu headroom till the input of the Apollo starts to clip. We did learn by wiz that we should leave at least 6 db headroom, and we did learn by svart that most brands use nicer looking numbers as they are. So I think you are safe with - 12 dbfs peak level in tracking. If the material has a super wide dynamic range you will even need more headroom.
|
|
|
Post by tonycamphd on Jan 12, 2016 23:17:11 GMT -6
Thanks mrholmes. The thing is, those specs didn't matter to me until I began to understand what's needed from all the helpful posts here. I basically understood to stay under -12 on my DAW's meter, which I of course ignored all too often. I'll try to be smarter about the process from now on, and ask for help here when I need it. I know the cats here won't let me down, or laugh at my questions, though I may cause a chuckle or two now and then. Correct me if I am wrong but given that you set - 20 dbfs equal to + 4 dbu it means you have 20 dbu headroom till the input of the Apollo starts to clip. We did learn by wiz that we should leave at least 6 db headroom, and we did learn by svart that most brands use nicer looking numbers as they are. So I think you are safe with - 12 dbfs peak level in tracking. If the material has a super wide dynamic range you will even need more headroom. Just peak average where they suggest in the manual, if you are tracking at 96k/32 float? as you should be for all of your processing and plugs to work and sound better, then you will be golden, you do get MORE HEADROOM using higher bit depths, i know wiz say's its foot room, and technically he's right, but i liken this analogy, standing on a concrete floor in your house, if you have 8' ceilings, and you jump and throw your arms in the air, you are bound to slam your hands into the ceiling(lets call that 16bit), if you go 24/32float, you've basically dug a 5' hole into your floor, and now your standing in it, when you jump and throw your hands in the air, you have little to no chance of hitting your hands on the ceiling, which means you now have headroom! haha see . You could also look at it like a window, if you increase the size of your window, and then drop the picture in the window closer to the bottom of the window, you now have more headroom above, ahaha.. see or.... um, I'm totally wrong, and i should be tarred and feathered 8)
|
|
|
Post by wiz on Jan 12, 2016 23:19:48 GMT -6
see on that graph, top is 24dBu, that is what you would see in the specs for the device, saying maximum input, line.
so, there would be 20dB difference between +4dBU (nominal 0VU point) and +24dBu , in this instance the maximum level it can take.
This means... in your DAW, when its hitting, -20dBFS, you are at line level.
If the top of that graph said +20dBu
with +4dBu being your line level and the nominal 0VU point.. you would have it reading -16dBFS on your DAW meters
both example above, assume your DAW meters are set to PEAK.
cheers
Wiz
|
|
|
Post by ericbradley on Jan 13, 2016 5:27:51 GMT -6
That's actually quite not completely true (or maybe i understood you wrong?) I tried to explain to the Mixbus guys at Harrison, too, because the dither is greyed out at 24bit bitdepth export in Mixbus, which i consider is a technical error, which is actually not good at all. While the audio is in the DAW, and only the slightest bit of processing is done, a dither will be needed at export to everything but the native format of the DAW. Which is nowadays 32 or 64 bit FLOAT, sometimes it can even be changed from one to the other like in Sonar. So if you write the file from the DAW to something like 24bit, or to a DAC, which also ends in an integer format, you NEED dither, because there is requantization going on. The format is always higher resolution in the DAWs float formats for processing. Even if you started with 16bit or even 12 or 8 bit recording files. Only exception would be, that you did not process the input file AT ALL. No gain, no format change, nothing, and going out with the same bit depth. In this case the use of a DAW is completely pointless - in and out identical. I guess there is so much confusion from the time when the internal DAW formats were integer, like the input files and the targets. There, things were a bit different. But well, we have 2016 now... Logic applies TPFD dither automatically when bouncing to 24-bit. There is no need to use Logic's additional dithering algorithms unless you are truncating to 16-bit.
|
|
|
Post by M57 on Jan 13, 2016 5:40:10 GMT -6
Logic applies TPFD dither automatically when bouncing to 24-bit. There is no need to use Logic's additional dithering algorithms unless you are truncating to 16-bit. So why is additional dithering available when bouncing to 24-bit? ..and assuming that it makes no sense to dither on top of dithering, does invoking one of the dithering options bypass TPFD? I'm not saying you're wrong, I have no clue about this stuff, but if TPFD is automatically applied, why is the default bounce setting "None." That would be misleading.
|
|
|
Post by mrholmes on Jan 13, 2016 8:17:07 GMT -6
That's actually quite not completely true (or maybe i understood you wrong?) I tried to explain to the Mixbus guys at Harrison, too, because the dither is greyed out at 24bit bitdepth export in Mixbus, which i consider is a technical error, which is actually not good at all. While the audio is in the DAW, and only the slightest bit of processing is done, a dither will be needed at export to everything but the native format of the DAW. Which is nowadays 32 or 64 bit FLOAT, sometimes it can even be changed from one to the other like in Sonar. So if you write the file from the DAW to something like 24bit, or to a DAC, which also ends in an integer format, you NEED dither, because there is requantization going on. The format is always higher resolution in the DAWs float formats for processing. Even if you started with 16bit or even 12 or 8 bit recording files. Only exception would be, that you did not process the input file AT ALL. No gain, no format change, nothing, and going out with the same bit depth. In this case the use of a DAW is completely pointless - in and out identical. I guess there is so much confusion from the time when the internal DAW formats were integer, like the input files and the targets. There, things were a bit different. But well, we have 2016 now... Logic applies TPFD dither automatically when bouncing to 24-bit. There is no need to use Logic's additional dithering algorithms unless you are truncating to 16-bit. Never heard about that but good to know. Here is another one about logic. Most people ask for PFL for leveling. Right logic has no PFL but it has in the transport bar the button PFM = Pre Fader Meteirng. Click that button and you can check you levels PFM.
|
|
|
Post by mrholmes on Jan 13, 2016 8:25:40 GMT -6
see on that graph, top is 24dBu, that is what you would see in the specs for the device, saying maximum input, line. so, there would be 20dB difference between +4dBU (nominal 0VU point) and +24dBu , in this instance the maximum level it can take. This means... in your DAW, when its hitting, -20dBFS, you are at line level. If the top of that graph said +20dBu with +4dBu being your line level and the nominal 0VU point.. you would have it reading -16dBFS on your DAW meters both example above, assume your DAW meters are set to PEAK. cheers Wiz And that is the reason why the RME has lesser headroom, sorry footroom because they say 0 dbfs is equal to 13 dbu. Input/Output level for 0 dBFS @ +4 dBu: +13 dBu 0 VU would be reading at the DAW Meter now circa - 28 dbfs and peak would be - 20 dbfs
|
|
|
Post by Martin John Butler on Jan 13, 2016 9:06:36 GMT -6
Thanks for that graph mrholmes. Maybe I'm a bit biased visually, but it's just a little clearer to me seeing that. Great posts guys, thanks for all of them.
My computer's long in the tooth, so I've been tracking at 48. Should I give 96 a go and see if I run out of steam?
Normally, I'd have bought a pimped out Mac Mini and the new Apollo with the satellite by now, but my business has changed so much, so quickly, I'm not quite sure what to do about it, so big purchases are on hold for now.
|
|
|
Post by mrholmes on Jan 13, 2016 10:07:36 GMT -6
Thanks for that graph mrholmes. Maybe I'm a bit biased visually, but it's just a little clearer to me seeing that. Great posts guys, thanks for all of them. My computer's long in the tooth, so I've been tracking at 48. Should I give 96 a go and see if I run out of steam? Normally, I'd have bought a pimped out Mac Mini and the new Apollo with the satellite by now, but my business has changed so much, so quickly, I'm not quite sure what to do about it, so big purchases are on hold for now. Its very easy if you keep in mind there is no real relationship between dbu and dbfs. It took me a while to get it, but there is a great German website explaining this. 0 VU is set different in the US than in Europe. US - 20 dbfs = 0 VU = +4 dbu EU - 18 dbfs = 0VU = +4 dbu Or something very wired. German Broadcastservice handbook says that files are not allowed to peak higher than -9 dbfs. Do not ask why I did try to google on that, but it just shows all the relationships between dbu and dbfs are approximations. I think wiz is right with knowing the specs of the AD. For example RME is not writing something about what the best peak level is. The RME sounds right to me with peaking around - 16 down to - 20 dbfs. You may read your manual and let us know if you find some information about it??
|
|
|
Post by Martin John Butler on Jan 13, 2016 13:09:56 GMT -6
Tony's post inspired me to try a 96k sample rate. I recorded a lovely acoustic guitar part after playing for an hour straight. Unfortunately, I keep getting System Overload with just a few of my must have plug-ins engaged.
I guess that tells me what I need to know, I REALLY need a new computer and more UAD dsp in general.
Call me crazy, but I could swear the high end was a little cleaner and more transparent at 96.
|
|
|
Post by jazznoise on Jan 13, 2016 13:48:01 GMT -6
You're crazy.
|
|
|
Post by Martin John Butler on Jan 13, 2016 13:56:08 GMT -6
Maybe it was because I tracked at much lower levels, because something was a little better. I'll do the same at 48k and see how that sounds.
|
|
|
Post by Ward on Jan 13, 2016 15:06:57 GMT -6
48k has always worked well for me. 96k works better for a lot of plugins. 88.2k, however, is a complete and utter waste of time.
|
|
|
Post by wiz on Jan 13, 2016 15:25:31 GMT -6
I have had audio interfaces, where it sounded better at 96Kz than 44.1Khz or 48Khz
I have interfaces where I hear no real significant difference between 96Kz and the other sample rates.
I have never bothered recording at sample rates, other than 44.1 , 48 and 96 Khz.
My advice... which as we all know, combined with 2 bucks, can buy you coffee.... 8)
When you get a new interface, after reading its specs and understand the levels side of things...
Do some test recordings at 44.1, 48 and 96.. then mix each of those down to a 44.1 Khz master.
If you are serious, get all 3 songs Mastered and delivered back to you as 44.1Khz 16 Bit files... and 320Kbps Mp3s
Then listen to the "result".
Then, you only have the processing power question to resolve.
If you hear/feel/see or intuit a difference that makes your mixing engineer brain, or your artist brain, glow and feel fuzzy... 8) use that.
If you have the processsing power, and the audio sounds good to you... use the best sounding sample rate to you.
cheers
Wiz
|
|
|
Post by wiz on Jan 13, 2016 15:30:54 GMT -6
see on that graph, top is 24dBu, that is what you would see in the specs for the device, saying maximum input, line. so, there would be 20dB difference between +4dBU (nominal 0VU point) and +24dBu , in this instance the maximum level it can take. This means... in your DAW, when its hitting, -20dBFS, you are at line level. If the top of that graph said +20dBu with +4dBu being your line level and the nominal 0VU point.. you would have it reading -16dBFS on your DAW meters both example above, assume your DAW meters are set to PEAK. cheers Wiz And that is the reason why the RME has lesser headroom, sorry footroom because they say 0 dbfs is equal to 13 dbu. Input/Output level for 0 dBFS @ +4 dBu: +13 dBu 0 VU would be reading at the DAW Meter now circa - 28 dbfs and peak would be - 20 dbfs A less confusing way to understand this..... Look at the spec for maximum input level for your audio interface. Eg +24dBu subratct 4 from it result +20dBu now that number is your 0Vu point. -20dBFS so if your audio interface was +22dBu, 0Vu equals -18dBFS if your interface is +19dBu, 0Vu equals -15dBFs maybe that makes the math easier... You buy your interface,... you read the spec.. you do the math.. then you do some test recordings.. keeping in mind the 0Vu point... over time you find where it works best.... cheers Wiz I hope this thread takes hold... cause mastering engineers will be sending me Xmas cards.... 8)
|
|
|
Post by mrholmes on Jan 13, 2016 17:16:26 GMT -6
And that is the reason why the RME has lesser headroom, sorry footroom because they say 0 dbfs is equal to 13 dbu. Input/Output level for 0 dBFS @ +4 dBu: +13 dBu 0 VU would be reading at the DAW Meter now circa - 28 dbfs and peak would be - 20 dbfs A less confusing way to understand this..... Look at the spec for maximum input level for your audio interface. Eg +24dBu subratct 4 from it result +20dBu now that number is your 0Vu point. -20dBFS so if your audio interface was +22dBu, 0Vu equals -18dBFS if your interface is +19dBu, 0Vu equals -15dBFs maybe that makes the math easier... You buy your interface,... you read the spec.. you do the math.. then you do some test recordings.. keeping in mind the 0Vu point... over time you find where it works best.... cheers Wiz I hope this thread takes hold... cause mastering engineers will be sending me Xmas cards.... 8) If that is true 0 VU on the RMEs would equal to - 9 dbfs and with this only 4 db headroom. 13 - 4 = 9 If I drive the RME like this it sounds stressd. Thats why I took the same proportion as with 20 db headroom. And it sounds good. If thats the way to calculate it the headroom of the RMEs is very small....
|
|
|
Post by wiz on Jan 13, 2016 20:35:09 GMT -6
It means that if the RME has a maximum input of 13dBu.... And nominal line level is +4dBu...... The 13 minus 4 is 9.
Therefore, if you send a signal at line level , it would register at -9dBFS on your daw or in total mix.
Which would mean you only had MAX 9 dB headroom above line level.
Make sense?
Cheer
Wiz
|
|