optimum mastering levels

[anurag]

I am just a beginner in the field of audio engineering, one question always lingered in my mind is about achieving best levels for my master. Though I gone through lots’ of articles & discussed around the same topic hell of times, but still puzzled.

Would like to seek your opinion & suggestion on following topics:

1. Fundamental understanding of levels
2. Mastering Levels on CD
3. Finalized Levels on Analogue Cassette Tape
4. Recommended Meters (Plugins) for identifying optimized levels

Kindly help me, if I wrote something stupid it was just due to my English ignorance. Please bear with that.

anu

 

[anurag]

no response

seems i wrote something very horrible........ well don know the protocols of this forum.

 

[Blue Bear Sound]

No - nothing horrible... but you've presented some pretty wide topics for discussion. Few people have time to write small novels on the broad question of "levels". (And the way you presented the topics, it would take a small novel to describe!)

I think you need to ask some more specific and pointed questions.....

 

[anurag]

ohhh! feeling pity

sorry guyz!

it would be great enough if you can specify my the ideal levels for CD Master. I heard that CD should contain peaks below 0, is that a standard? How to identify optimum levels for analouge & digital media, any specific plugin?

luv

 

[Massive Master]

Any peak beyond 0dBfs is a CLIP.

Clips are BAD things. A good brickwall limiter will stop digital clips from dynamic material IF used properly. Even the best brickwall limiter starts to sound horrible if the signal is pushed into it for the sake of sheer volume.

During tracking, I try to keep my hottest level below -6dBfs. In 24-bit, there's no reason to want hotter.

During mixing, same thing. -6 or -4dBfs PEAKS, with a crest factor of about 16-18dB depending on the style of music.

During mastering, the maximum peak should still be BELOW 0dBfs. My personal favorite is -0.3dB at the absolute max.

I hope that helps...

 

[JKestle]

differece between 16 & 24 bit....

I have no choice but to record in 16 bit right now.

Massive Master, I think I understand why -6dbFS is a good level for peaks in a 24 bit session, becoause the noise floor is much lower compared to 16 bit right?

So my questions is, when tracking in 16 bit for the time being, what am I taking away from my recording by tracking closer to -3 to -2 dbFS on average?

Follw-Up (if you have the time): is there a significant advantage to upsampling to 24 bit for mixing 16 bit tracks?


thanks a million, hope this helps Anurag as well.

 

[Massive Master]

Yeah, 16-bit you want to get fairly hot. -3 or -2 is puh-lenty.

Depending on your software, *most* audio software is working internally at 24 bit or better anyway, so the actual mix won't make much of a difference.

HOWEVER (as usual) - If you're sending out for mastering or doing any other processing at all, mix TO 24-bit files so you're not adding noise or dither artifacts that will have to be added during the final step.

 

[JKestle]

internal mixing eh?

That makes sense, great...thanks!!!

 

[lpdeluxe]

...dither artifacts that will have to be added during the final step.

Aha! So THAT's what mastering engineers do!

 

[karyoker]

In sonar3 I've developed sort of a sense for levels & try to track -6dbfs.

But if you are tracking or mixing with sonar3 or equivalent how accurate are the meters? Is this what you use for levels? If so what db range do you use? If changing from 16 to 24 bit with a broader range do you change the db range? How accurate are the peak to peak scope views in the track panes? Some have a db scale on the start of the track.

I've been tempted to drag out an old scope and actually see whats going into the breakoutbox That would be analogue b4 the adc but I sort of rely on the digital meters....They seem to work....

 

[masteringhouse]

I have no choice but to record in 16 bit right now.

Massive Master, I think I understand why -6dbFS is a good level for peaks in a 24 bit session, becoause the noise floor is much lower compared to 16 bit right?

So my questions is, when tracking in 16 bit for the time being, what am I taking away from my recording by tracking closer to -3 to -2 dbFS on average?

Follw-Up (if you have the time): is there a significant advantage to upsampling to 24 bit for mixing 16 bit tracks?

Since 6 dBFS is roughly equivalent to 1 bit, peaking at -6 for a 16 bit recording is like recording at 15 bits, while recording at 24 bit is like recording at 23 bits, still much better than even 16 bit even at full strength.

To me recording tracks is a different story than a final mix. Due to the summing bus you'll want to be more conservative with tracks than a final mix. In general go hotter with a mix than when recording a track. -3 to -2 dBFS shouldn't be a problem in a mix.

You are not "upsampling" when going from 16 to 24 bits, upsampling would be when going from 44.1K to 48K or higher for example (higher sample rate). What you are doing is increasing the bit depth or word length. This is a good thing since it will help to reduce quantization error when processing in 24 bit and then going back down to 16. There are also advantages in upsampling (going to a higher sample rate) in regards to aliasing assuming that you have a good sample rate converter.

 

[JKestle]

To me recording tracks is a different story than a final mix. Due to the summing bus you'll want to be more conservative with tracks than a final mix. In general go hotter with a mix than when recording a track. -3 to -2 dBFS shouldn't be a problem in a mix.

I'm sorry to ask, but could you explain the summing bus, I'm not sure how it relates to conserving tracking level.

You are not "upsampling" when going from 16 to 24 bits, upsampling would be when going from 44.1K to 48K or higher for example (higher sample rate). What you are doing is increasing the bit depth or word length.

oops, thanks for the correction.

 

[anurag]

ahh kewl!

i majorly tracking in 24 bit using Nuendo. I think tracking on high bit rate gives us more headroom, is that so? One more question, what will be the ideal levels for master on analogue tape ?

 

[masteringhouse]

i majorly tracking in 24 bit using Nuendo. I think tracking on high bit rate gives us more headroom, is that so? One more question, what will be the ideal levels for master on analogue tape ?

Sorry if I'm being pedantic, but not "bit rate" its "bit depth". Yes more headroom, with 32 bit floating point being the most forgiving v.s. 16 or 24 bit fixed.

The ideal levels for analog tape really depend on the sound that you want to acheive. One of my old engineering friends used to say "If you're not in the red, you're dead". Basically meaning that he wanted that tape compression sound for things like guitars and drums. So for that classic rock sound you'll want to go hotter than if you were recording Jazz or acoustic music. Best thing is to compare the sound of the source angainst the repro head to see how you're affecting the signal.

Analog is kind of the reverse of digital this way, for tracks you'll probably want to go with a hotter level to help with the signal to noise ratio, and for tape compression (in rock), while the mix won't be overly aggressive in order to maintain the overall sound. For mixes go for 0 on the VU meter and keep the peak light from blinking too much if at all.

 

[masteringhouse]

I'm sorry to ask, but could you explain the summing bus, I'm not sure how it relates to conserving tracking level.

Quite a few threads on this already here, so I won't rehash all of it. Basically a summing bus adds all of the output from the individual tracks to produce the stereo (5.1, etc.) output. If the output of all of the tracks exceeds 0dBFS you'll get clipping and need to lower all of the tracks by the same amount (if you want to preserve relative levels) in order to get the sum below 0dBFS. By recording at levels that are close to the level that you want in the final mix, you will hopefully not need to process the volumes of the individual tracks as much thereby reducing the additional artifacts of digital processing for those tracks.

 

[anurag]

yuck!

i mixed sample rate with bit depth........ i just read about bit depth "audio frozen in a single moment of time" ..... is it right? What more you can expect from a newbie...........

thanks for great suggestions........

luv
anu

 

[masteringhouse]

i mixed sample rate with bit depth........ i just read about bit depth "audio frozen in a single moment of time" ..... is it right? What more you can expect from a newbie...........

thanks for great suggestions........

luv
anu

Bit depth (16, 24, etc.) is what controls the amount of headroom of a system (amplitude at a particular time), sample rate controls the frequency response in the most basic terms.

 

[lpdeluxe]

Sample rate vs bit depth

Sample rate is simply how often the analog signal is sampled. Standard is 44100Xsec for CDs.

Bit depth is the number of possible levels of amplitude of any given sample. The higher the bit depth, all things being equal, the smoother reverb decays (for example) will sound. Lower bit depth leads to noticeable steps as the ADC tries to match the level of the incoming sample with its pre-programmed levels as determined by its bit depth. Think of digital audio as being wave forms that constantly change being converted into slices of those wave forms, with the tops of the slices cut off at preset points. In other words, you are taking a continuous wave and making it a stairstep structure. 16 bit looks more like the original wave than 8 bit, and 24 bit looks even more like it, i.e., smoother. It's like a loaf of bread before it's been sliced versus after; you can never really reconstruct the exact shape of the loaf from the slices, but the thinner the slices (sampling rate) and the higher number of possible lengths of each slice (bit depth), the closer you can get.

 

[masteringhouse]

lp -

Don't forget about smoothing filters on the DAC. The output really isn't a stair-step going from digital back to analog.

 

[JKestle]

Quite a few threads on this already here, so I won't rehash all of it.

sorry about that, the terminology threw me off.

Basically a summing bus adds all of the output from the individual tracks to produce the stereo (5.1, etc.) output. If the output of all of the tracks exceeds 0dBFS you'll get clipping and need to lower all of the tracks by the same amount (if you want to preserve relative levels) in order to get the sum below 0dBFS. By recording at levels that are close to the level that you want in the final mix, you will hopefully not need to process the volumes of the individual tracks as much thereby reducing the additional artifacts of digital processing for those tracks.[/QUOTE]

OK, this is clear in my mind now. thanks for your help.