Vocals - Signal:Noise - Principles

can't really answer your question with the information at hand, but I can give you some ideas...
 
Back in the mid 1990s (I think) Louis Fielder, one of the engineers at Dolby Labs, wrote a landmark paper titled something like Dynamic Range in Digital Systems. In this paper he suggested that the proper reproduction of a musical performance could require as much as 120 dB of dynamic range. That's a BIG number, a ratio of 1,000,000,000,000:1 (yes, that's 10^12 - feel free to check my math!)
 
That is theoretically possible, and while some chips claim that dynamic range, it is more of a marketing number than a real number.
 
Let's think about that for a minute... a really quiet recording space would be built to have a noise floor on the order of NC20, or about 20 dB-SPL broadband, average. And it is generally accepted that the threshold of pain occurs somewhere around 130 dB-SPL (depending on age, physical condition, and the content of the noise.) So it would seem that we are already in trouble<G>!

So what's practical? Or realistic?
 
You have to start with the room! How quiet is it in your room. You'll need to borrow (or buy) a real sound pressure meter if you want to know the answer, but you can get by with your recording gear if all you really care about is the S/N ratio in your recording.
 
But whatever you end up with is, well, what you end up with. The noise in the room is the noise floor, it's only going to get worse as you pass it through transducers and electronics, and especially modern (read low power) A/D converters. For sake of argument, lets say that the average SPL in your recording space, across the entire audio band, is around 30 dB-SPL. So even if you could hit (and survive?) 130 dB-SPL you'd still only have 100 dB of dynamic range... IF none of the subsequent stages distorted or created noise of their own... and they all add noise!
 
Practical... gotta remember that word!
 
I usually monitor so that the wide band, average level (slow response) is around 85 dB-SPL. Remember, this is what is delivered to my ears, this still has nothing to do with the electronics, that's a whole 'nother matter! And we, by which I mean "I" keep throwing that word average in...
 
It turns out that the human ear is sensitive to both average and peak levels. And that is one of the places where people tend to misunderstand poor Louis's assertion. He was thinking about the whole kettle of fish, and especially peaks. Peaks take on a couple of forms in music. There is a peak when you strike a snare drum that is very short in duration, but depending on the drummer and the drum, it can be REALLY LOUD for a fraction of a second.
 
And that' what Fielder was trying to preserve. Well that, and the natural "crest factor" of complex waveforms. And that's what you need to consider as you set up your system.
 
We really don't know much about the energy in the noise floor you are measuring in Sonar. We don't know who the contributors are, or what they contribute, but since you probably can't do much about it, and 90 dB down for an average noise level is quite decent really, let's forget about it, and worry instead about the upper end.
 
You probably know that you can not exceed 0 dBFS - which is the maximum level that the hardware and/or software can represent. What you may not know is that level is going to be a different physical level at different places in the system, and that can be a problem. You should also keep in mind that ALL dB measurements represent power ratios, even though we use them to represent sound pressure levels and voltages and whatever. And this means that they are, by definition, average or RMS (root-mean-squared) measurements. They are not peak measurements, and they can't be peak measurements.
 
The trick then, is to find a spot somewhere below 0 dBFS where you can operate safely, and by safely I mean:
1) you never exceed 0 dBFS for more than a couple of samples
2) you don't clip analog stages before or after the computer.
 
Some folks think that you should treat a digital system the same way we treated analog tape. That's not a good idea. With analog tape we had a huge noise contribution from the tape, and probably the tape electronics. So we tried to operate as close to 0 VU (an entirely different measurement system) as we could. We don't need to do that any more. Digital storage and transmission do not add noise to the signal, only the channel, and the channel gets tossed away!
 
So unless you are recording the song of the Egyptian Fruit Fly (and since you mentioned amplifier hiss you probably aren't), the key here is to find a level in the digital domain that works for you. You need some headroom to accommodate peaks, and you want to stay as far away from the noise floor as possible... without exceeding 0 dBFS or clipping.
 
Sensing a pattern here<G>...
 
I tossed a fair bit of information that you may or may not need, but I hope it demonstrates the futility of sweating S/N ratios, or even dynamic range, in a modern recording environment. The bottom line is that the noise that is inherent to your recording chain is going to be there. If you record your tracks at a level significantly higher than the noise floor then the noise will still be there, but listeners probably won't hear it.
 
One last thought - a lot of this is pschological, or psycho-acoustic. When I was still working with tape we would use a trick called spot-erasing, where we would erase the noise in between the notes - ok, seldom got that silly, but if the guitar amp was humming a bit we'd erase the tape if the guitar player wasn't playing. It tricked the mind into thinking that the dynamic range of the PERFORMANCE was greater than it really was. (When inexpensive automation became popular we'd automate the mutes to do the same thing.)
 
Have fun... and don't spend too much time worrying about S/N ratio... or reading this post for that matter!
 
 

If it makes you feel better I don't think you are alone in overcompensating for noise. When I had a "real" studio it was so much easier, but for now I record in the basement, which isn't awful, but it sure isn't quiet either. I try to focus on the fact that the room itself sounds nice when I'm feeling positive, and I try to use microphone selection and placement to minimize the noise problems.
 
Anyway, if I do wander off into that dark spiral of noise reduction after the fact I tend to end up with a mix that sounds awful... usually better to either leave well enough alone, or place the microphone close to the source!!!
 
Suggestions... geez, guitar and vocal... well Richard Thompson immediately springs to mind, his solo stuff is brilliant - all his stuff is brilliant really. Leo Kottke would be another, especially his really early stuff, there is noise there, but you won't notice it. Paul Simon, James Taylor, Dan Fogelberg, Jackson Brown, Lou Reed (don't laugh, he did some stellar acoustic guitar and vocal tracks), Warren Zevon, Townes Van Zandt, Rodney Crowell, Guy Clark, Steve Earl, Jimmy Dale Gilmore, John Hiatt, Lyle Lovett, Vince Gill, Frank Christian, Joe Ely, Robert Earl Keen, James McMurtry, and Tom Waits (his recordings of his demos are often guitar and vocal, they are a bit rough as they were never intended to be published, but still quite cool.)
 
And that's just the male singer songwriters that did solo guitar/vocal tracks that I can think of off the top of my head... there are lots more. And they cover the gamut of styles...
 
An acceptable noise floor is very much a function of the setting. My last studio was down in the NC30 range, which is barely good enough for a commercial facility, and yet now I'd love to be able to get there<G>! So it's also relative I guess...
 
One way to look at it, the way I look at it these days anyway, is that if the listener is bothered by the noise then I goofed. Either the performance is not sufficiently good to keep their attention, or there is a real noise problem (or both I guess.) Once you enter the D/A converter there is really very little opportunity for contamination, and it stays that way until you exit the D/A converter. And from D/A to ears is HUGELY variable and beyond your control.
 
That means you just need to focus on the room, the microphone selection and placement, the instrument (which includes amplifiers) the preamplifier, any processing prior to the A/D, and of course gain staging. Of all of this gain staging is probably the most critical, followed by microphone placement. The rest really should not play into the equation, unless the room is really noisy - at which point we repeat the bit about microphone placement<G>!
 
Does that help a little?

The SM-7 is an industry standard, but needs a lot of gain.  The ISA should supply that, but that is covered in your other post.  Realistically, both your  level seems a bit low, unless you have a quiet room so you can add gain in the digital realm.  The ISA has more gain available.  I'd try that, +5 or 10 dB.  For such a simple (kinda like dumb questions - there are no dumb question but simple can devolve into complex quickly) project, you should have lots of headroom mixing a vocal and guitar at those levels.  I've found it easiest to lower gain than try to add it after it goes digital, as long as the recording is good.
 
@

In my mix I am keeping my vocal track level at 0 gain and mixing around that. I use an expander to bring the noise floor down a bit while maintaining the same output level as the source.
 
I bring the acoustic guitar part way down until the vocals sit on top and I adjust EQ and stereo spreads to give the vocals and guitar their own space in the mix.
 
On my master bus is some light compression to bring the overall level up a bit but this tends to undo the noise reduction from the expander in my vocal track. I try to balance all these things until I get something acceptable, and I think I have, from a technical standpoint, now I am working on getting it to sound "pleasant". 
 
Does this sound like a good start?
 
(edited first paragraph for clarity)

Thank you for all your help. It has been very informative.

Bitflipper has it above. I'd add this: why are you using an expander? Never use anything without a reason.
 
One really good rule of thumb, in all audio engineering, not just this, is to see how little you can do. 70db is a really pretty good signal to noise ratio. Maybe you don't need to do anything much.
 
If you do want to lose noise, you'd be better off just slip editing out the silence, rather than use an expander. They're kind of brute force, last resort tools, really.

Hmm.
 
Here's my suggestion: try doing a couple of mixes without trying to get rid of the hiss. Then live with them for a bit, and see what you think. Lots of great records have a bit of hiss on them. Indeed, it can be kind of soothing to the ear, in the right context.

I don't know this for sure, not having heard the material, but from what you're saying, I think you might being having a "not seeing the wood for the trees" problem here. Do you actually have problematic hiss? At the SNR numbers you quoted, I wouldn't think so.

I just bought my first ever condenser mic, and the learning experience of it was interesting. I actually live in the country, so there is very little ambient noise, but the microphone was so sensitive it was driving me batty. I tried everything to lower ambient as much as possible EQ/gating/comp thresholds, you name it, but I found the solution (that suited me) was a simpler answer.
 
I really didn't participate in the forums much until X3, so been learning a lot recently, and posts about noise reduction with Audition/Audacity have passed by. I own Audition (I believe Audacity is free), but never really explored it until recently, and it is useful for this...
 

1. When recording audio, I will let a good 5 seconds of pure ambient noise run at the beginning AND end of a track, do my recording, save, and close X3 (so X3 "releases" the audio file allowing it to be edited/saved elsewhere).
2. Open that track in Audition, which can first be used to "declare" the ambient noise (I choose the tail of my choice), and then with the clip highlighted, can then "process" the clip to remove that ambient from the track. Save, close Audition, and reopen X3.
3. When the project reloads, the tracks are incredibly clear and crisp, so this "work flow" has become standard for me now. After the "raw" is cleaned, I use X3 to do the rest.

   I have not had to get into what Audition is capable of (blindly declaring ambient and processing the clip has worked fine), and from posts I have read, Audacity is recommended by many for use in a similar way.

Noise is good... Just use envelopes that cut it out on the intro and the outro. Then when you hear it "In the mix?" You won't hear it.