I had posted a long reply to Jake, but broke off to watch the Transit of V, and when I returned my post had gone. So this is a summary!
I've used Audacity before, but its MP3-export is new to me. It requires first installing LAME, which I did to the letter of Audacity's instructions. Exporting MP3 then makes NO provision for tweaking of any sort, you take what you get and like it. Well I don't, neither do you. I'm in the market for an editor which does MP3 better.
I rank the quality I hear like this, where the chevrons indicate degradation : >N3, >>>32bit WAV export from N3, >>>>>>LAME MP3 export from Audacity. So MP3 is twice as bad as WAV, probably more. N3 scores a chevron simply because it can't be perfect, but it's the best I hear, by those degrees of chevron-differences from the other two steps. And why N3's own 32-bit WAV export (!) should be that much degraded beats me.
I'd previously begun a description of N3's Mixer, which etto calls "excellent", and I wouldn't know, no comparisons to judge by. It has 3 sorts of entity sporting controls : Channel Strips (as many as you have staves); Buses (A, B, C, D); and Master. Channels can output to Master, to a single Bus, or to a mix of Bus and Master where the ratio is set by a little twiddler inside each channel strip. The default is 0% to Bus, 100% to Master. A third option is thus 100% to Bus (and THEN either to Master or external equipment). The external output option also holds for each channel, and consists of UP TO 32 stereo-pairs of outlets, you can implement how many and which of the 32 from inside each channel, so you could have at saturation 32 staves each going a different place, and above that (say above 36 staves) some amount of mixed-output would be needed. As well, you might WANT mixed-output to external inputs, say you had just 4 external inputs to go to, and 16 instruments going inside N3.
Since output-default is 100% to Master, it is Master-output I hear from inside N3, and which >N3 uses to export its 32-bit >>>WAVs from. So my my best hope of nixing both > and >>> lies in some now-unknown software (probably VST) to slot into the external-input provision Mixer gives me. And I'm looking for what might be out there right Now.
AT the same time I last posted on THIS topic I was also downloading the 7 GB of sound that Miroslav Philharmonik has. Yep, I got scared into saving ~30 Euros on a deal that had one day left, so I jumped eyes shut. When I dropped Miroslav Strings into the slots of the N3 Strings 'playing' in PRAYER 1B.MP3 as posted here, I was at first DUMBSTRUCK, so florid the result was by contrast. That effect has since worn off, and the Miro strings now sound ordinary.
Like etto I'm no expert at tweaking Miroslav, my Miro-use has been confined to what N3 gives me, just notating, a bit of reverb, and Mixing and Panning. At its right-hand end, N3's notation palette has a tab that expands into a list of techniques the N3 instrument you have currently-selected offers, its effects. That's true of Miroslav instrument effects too, and it looks to be a bigger list, but not by very much. The floridity is the main diff between sets.
Indeed, to answer one of your questions directly (forget the number), it could well be a relative monotony of the N3 default samples NOTION counts on to sell its London Symphony offering. THAT could be the "unknown setting" you suspect is present. (Not but what such a setting might exist, unknown to me; I'm STILL looking for a command that'll let me add pages to a score, without luck. Surely one must exist?)
Plus the impression of floridity wears off. That'd bear repeating. What you hear gets 'ordinary', fast.
Anyway I'll now re-post the N3 score of PRAYER 1B with Miroslav dropped in where N3 was, as PRAYER 1J MIRO.MP3. Also in Mixer the locations of the strings and Pianoteq is tightened up. Through headphones (not marvelous ones either) the hearing's much sharper. Of course, >, >>>, and >>>>>> still apply, but Miroslav plus the tightening (and headphones) must help.
EDIT: CAPITALISE a sentence-start.
ADDED : Just noticed. It is not "crunching of the the volume levels" I meant. If I said it I misspoke, and can't recall doing so. No, it is DEFINITION that has gone, THAT's the degradation, and why is not a wonder. You can't throw away data in pursuit of compression, and expect anything other than relative BLUR to result, the more you ask compression the worse the blur.
I suspect that Audacity's use of LAME to export MP3, paying off in a dialog that doesn't allow selective compression (like I *know* other MP3-crunchers do) results in a savage compression, ie blur. One WAV I started with (N3 is at 32-bits in my hands, remember) was ~17 MB, while after LAME was through, it was ~500 KB.
One thinks of shrunken heads. Off now to crunch PRAYER 1J MIRO.
MORE : upload done, and the MP3 you hear is A) a crunch by SOUND FORGE AUDIO (I just couldn't stand the screech-and-boom of the Audacity/LAME version); B) much smoother as to screech, even via Mylar headphones. If you have an equaliser in your sound chain, you might want to drop the Bass a bit. Plus I now realise I boosted the Pianoteq part too much. Makes the Strings a mere accompaniment.
Last edited by custral (06-06-2012 18:20)