how to avoid clipping?

I think you need to reduce your input volume as a starting point. I find with Audacity that once I've imported everything I can get some decent results using Select All and the Normalisation function, and then dealing with channel volume by use of the Amplify function and a bit of gain control.

It doesn't always give me exactly what I'm after - I put a song on my Reverbnation page today which still has some hiss on it, but overall I find the results are ok.

A :-)

Hiya, Jason...a few points. I'm not quite sure of what you mean. Clipping? You're not using tape, are you? Separating out the background noise from a distorted sound is defeating the purpose of "the distorted sound"...or are you wondering how to sort out the unintended distorted sound??? Like I said, I'm not real sure of what you want to accomplish. As for MP3: it stinks...you cannot establish a high enough Nyquist Frequency to cope with the dynamics. And, your point of having to spend thousands on the right equipment, sadly, is the plain fact. Sure, El Cheapo will work just fine...if you want a dirty kind of sound. Pro software is just that...pro. But it costs, heaps! Fact of life, I'm afraid!

Cat

Ouch, I'm afraid I have to disagree with practically everything Cat just said. Background noise (computer fan, streetnoise, self-noise etc) has nothing to do with distortion and should be filtered out as good as possible. The relation between Mp3, nyquist frequency and dynamics is nonsense, really. I think Greg already explained it, if not Wiki will help. The problem described doesnt require 'pro' equipment at all.

Anyway: the mobile pre and that MXL are certainly not top-of-the-line but more then enough to get decent recordings without that problem. When recording highly dynamic sound I suggest using a limiter, even if it's a dirt-cheap Behringer. For the rest just lower the input-volume (which will increase noise!), add a software gate and compress the remainder. Always try to send in the signal as loud as you can without clipping to reduce S/N. As for MP3, make sure you go for at least 192kbps, preferably more. 128kbps (which the free soundclick accounts use) is pretty hopeless and will not just reduce dynamics but add artifacts, reduce frequency range and often enhance S/N. Box.net allows 320kbps for free accounts, try that.

Also, audacity sucks balls for mixing. Try Kristal for mixing, it has a build-in compressor/limiter and it's free. Didn't your Mobile-pre come with Cubase LE or something?

it came with something, can't remember what except that it was so inconvenient that i went back to audacity. i don't do a whole lot of mixing. mostly just one off demo recordings so i can remember how my songs go, but i also post them from time to time and everyone tends to notice the noise before anything else, or else the volume is too low.

I read an article somewhere (sorry, don't remember where) that said you should try to keep your levels below about -18dbs when recording. That is the "0db" mark in analog recording. I used to make it as loud as possible without clipping, but when you combine several tracks, they end up clipping because they are all maxed out. You can get a lot more dynamics keeping the incoming signal below the -18 mark. To get it, I have to turn down the gain on my mic quite a bit. It does seem to be resulting in a better sound.

JWmartin: The lower you record the louder the native noise will be. To prevent the loss of dynamics you';re talking about it's better to record as loud as possible and then lower the volume of the track in the mixing program.That way you get low noise and all the dynamic space you might want.

Ignar: I'm not following what you are saying. By native noise, do you mean noise picked up by the mic? By reducing the gain and volume of the mic, it will reduce the background noise. I didn't mean turning down amps, I actually turned mine up and turned down the gain on the mic. Before, I could hear all kinds of ambient noise in my recordings, now I don't.

It's not about ambient noise, or noise the mic physically picks up. It's the noise created by the pre-amp itself and such: noise that is there regardless of the volume of the pre-amp. The higher the volume the lower this self-noise relatively is: increasing the volume post-recording will increase both the recording/ambient noise but also the self-noise. As this self-noise is relatively fixed sending it as loud as possible to the pc/recorder (relatively low self-noise) and then decreasing volume (lowering both sound and noise, keeping ratio the same) will result in lower perceived noise.

So in short:

1) send signal in as hot as possible for best S/N ratio.
2) adjust volume in the mix to get the required dynamics.

OK, gotcha. I thought you were talking ambient noise, but you meant the electronic hum inherent in the system. Makes sense.

So in short:

1) send signal in as hot as possible for best S/N ratio.
2) adjust volume in the mix to get the required dynamics.

This is exactly correct -- for each stage at recording (mic, preamp, DAC), maximize the output from the previous stage to not-quite-clipping the next stage. Always assume that once a linearly recorded S/N ratio is established in a recorded track, you will never be able to improve it in that native recording, so S/N must be as good as possible when created (let's avoid DSP post-processing noise reduction discussions here -- too many compromises anyway). so as Arjen writes, record each track as hot as possible without clipping. later on, when you reduce the level of each track at the mixing stage, you will reduce both the signal level and the noises carried along during the track recording (ambient, mic noise, preamp noise). Hopefully the noise inherent in the mixdown system is always significantly lower than that carried within the incoming tracks.

In linear recording, as in all linear communication systems the rules for producing the best S/N are

1. maximize the input signals levels to each stage.
2. make sure the up front devices (mic and preamp here) have the best noise performance (lowest noise figures), as noise from these stages will be carried through all subsequent stages to the final device.
3. put the bulk of the gain distribution toward the front end of the processing (mic, preamp).
4. put the bulk of low gain and attenuation at the back end (mixer) -- as that will have little effect on source track S/N until recorded track noise drops below the self noise of the back end (mixer)
5. make sure the back end (mixer) has a very large dynamic range.

why this is all true may not be entirely obvious, but becomes clear with a couple calculation example. I'm too lazy for that right now.

for working with a lossy codec (MP3) recordings, a similar philosophy works well: stay as high quality as possible for as far through the process as possible. and do the final conversions to MP3 or lower rate MP3 after mixdown. if I'm given a track in MP3 track to use with my work, I prefer to convert it to a linear PCM (ex, WAV) file and do all my recording in linear PCM as well. after the final linear mixdown to a linear PCM file, I will convert that to the various MP3 versions needed. this helps minimize nasty artifact generation. converting the original MP3 track to a linear PCM does not make it sound any better, but it does help minimize further corruptions to that track in the various FX and mixing processes.
OK, gotcha. I thought you were talking ambient noise, but you meant the electronic hum inherent in the system. Makes sense.

careful ... hum and noise are two different things. far easier to remove single-freq or harmonically related hum from a recorded track than noise (think hissssss).

right, and that's where the problem is. i try to record as loud as i can to avoid noise, but then i run into distortion from going over the red line.

and which stage is showing the clipping? Audacity? Preamp/PC interface metering/config utility (assuming there is one)?

both, actually, although the preamp clipping isn't as much of a problem for the sound quality.

can you lower the output level (not input trim) on the m-audio preamp/interface? or is that when you run into noise issues? then ...

I don't use Audacity. does it have a bit depth setting? that would change the dynamic range ... about 6 dB per add'l bit.

also, are you working completely in MP3? if so, that should have a bit depth (8, 16 ..) setting in addition to bit rate and sampling frequency. usually the anti-aliasing filtering is automatically set with the sampling freq, so for a fixed MP3 bit rate (ex, 320 kb/s) you can trade freq response and bit depth (dynamic range) for coding quality (magical mix in the MP3 algorithm). coding quality is not S/N, but levels of artifacts -- slushiness, phase distortion, warble ...