I make acoustic recordings with a (too) wide dynamics. They really need to be compressed for more realism.

Our ears will modulate the loudness but listening to an accurate recording of a wide dynamic doesn't have the same effect on the ears. Things turn out to be too soft or too loud whereas they were ok in an open space realtime situation.

Is there any definitive good practice audio level compression methods?

Thanks

These people have a free downloadable mastering guide that is supposed to be very good. It is oriented to their own software but a great deal of the information can be translated to any program with similar tools.
http://www.izotope.com/

http://www.digido.com/portal/pmodule_id=11...der_page_id=18/ and Bob Katz book as a start possibly.

I have a similar need and I finally settled on WavePad. It is a free general purpose sound editor. Available here:
http://nch.com.au/wavepad/index.html

Just use the Automatic Gain Control (AGC). I found it to be exactly what I needed.

I also noticed that the current edition of GoldWave has an AGC function. I did not care for it, but take a look. You might like it.

Woody

I think what the OP is looking for is a multiband dynamics compressor.

I don't think I need multiband compressor and I doubt if AGC would work though I will investigate. I haven't looked but the izotope work may be helpful. The Katz book looks too general.

I really need to know ideal attack and release settings for compressing the dynamics enough on my accoustic classical venues so the listener doesn't have to keep turning the vol up and down. Classical music composers relished dynamics in their works but I don't think they record realistically unless you can playback with big volume and in a huge room to abosorb the 0db stuff and still focus on the -24db stuff.

THanks all for replies

There is no one good setting for acoustic guitar recording/post processing. The really nice acoustic guitar recordings rely on miking technique, room, quality of the instrument and player's skill a lot. Especially the last factor is huge. I recorded for living for quite a few years and learned that there is no substitute for performer's skill and talent. Sometimes just turning the mics on and bringing up the faders on the console resulted in woderful sound. Sometimes I would keep fighting the equipment and mic placing for 20 minutes and still not be happy.
Anyway, here a good method for capturing natural acoustic guitar sound that worked for me on many ocasions.

First, you should record to "tape" through two microphones (one up close and one above the guitar maybe 18inches back and pointing at the sound hole ) and compress both with relaxed, slow settings, with around 4:1 ratio so that gain reduction in loud parts reaches maybe 6dB. Unlink the compressor channels for more natural and dominant room sound.
I liked the JoeMeek compressor for acoustic guitars going to tape a lot.
If you have EQ, patch it before the compressor, roll off the low frequencies below 80 Hz and apply a bit of low shelf reduction on the close mic. Each mic sounds a bit different so I can't give you a set value. You want the proximity effect to be neutralized. Too much low end and your guitar will sound muddy and work the compressor in a wrong way. It's easier to get the low end just right after compression is dialed in.

Once in the computer, split your stereo recording into two mono tracks (unless you captured them as two mono tracks already) and zoom in on the waves. Slide one of them until you align the attacks of sounds as close as you can. Pan them at 10 and 2 o-clock and listen for comb filering effects. If it sounds flangy slide the tracks a bit more until you hear them come in focus. Adjust the balance between them to taste.
Now you will want to dial in a bit more aggressive digital compression with just enough attack time as not to dull the fast sounds (1-2 msec) and short release time. The gain reduction should not be more than 6dB in absolute loudest peaks and hit maybe 2-3dB on average.
If this still doesn't sound quite right you might want to manually turn down loud parts and gently boost quiet parts by drawing a volume curve on the tracks. Make the transitions complement the music.
Give the output a bit of ducking reverb. For slower tunes use longer decays, for faster, busy tunes use shorter decays. Process in place and normalize the output. Now you should have a pleasant recording that still has some dynamics but doesn't sound weak.
It's a bit of work, but that's what it takes to get a nice sound.
You might also read some posts at recording message boards such as the prosoundweb.com.

I would say that there is no true formula, just try and trust your ears. Reading stuff mentioned above will help too.

You should watch out if you use Goldwave. If "Anticipate attack" is ticked on it will use the attack as 'look ahead' instead of the usual way. "Use Smoother" seems to do something unusual too, so you should tick them off if you want it to behave like a usual compressor.

You might not need a compressor at all if you record on multiple *forgot a word*

If you record on a single *same word* a multiband compressor could be usefull. I think goldwave can handle vst-plug-ins, and there might be some free multiband compressor plug-ins out there somewhere, if that is what you need.

And remember, as mentioned above, the performer is the most important part of the story.

Tracks?

I'm not sure what effect if any the anticipate attack has, but the smoother definitely minimizes the clipped tops, and preserves the shape of the wave form, at least when the attack and release times are 0. But when attack is set even with small amounts, the wave form is preserved and the smoother becomes unnecessary.

I was under the impression that I want to set the threshold high and the ratio steep and attack times low to gain head room. But after reading more I may want to look at making the ratio mild and the threshold lower. That may be less intrusive.

It definitely takes work and experimenting.

Multiband is not a good idea, unless you have a tonal ballance problem or certain frequencies out of control (sss and pop on the voice for example).
Unless I have a very good outboard compressor, I prefer to record everything dry in 24 bits with at least 6db of headromm and choose a compressor during the mix. Substractive equalisation (low pass, high pass, roll off, ...) is often better before the compressor unless you want to pump the compressor. Waves compressor (Rvox, Rcomp, Raxx, ...) are good but expensive. It's easy to find a version you can try (before you buy;). Sonalksis and Awida are good. You have also free plugin. You can find them here:
http://userpage.fu-berlin.de/~thman/VST/vst-alleff.htm
http://www.kvraudio.com/

Don't overdo it. Not more than 4:1 except for the voice (sometimes two compressor are used).

Cheers,