indeedsir_backup | (no subject)

I hope it's acceptable to post this here, as it's not fic but is *about* fic...

The background:
I was asked to podfic my little offering and I did, but I really wasn't happy with the result, which is why I've not posted it. I know part of my dissatisfaction with it is that one of my first jobs was in a recording studio that specialised in speech, and I got used to having a clean signal and some very good software to edit with. Currently I have my laptop's built-in mic and Audacity. It seems like a good piece of software, with most of the functions I had at work, but I can't satisfactorily remove the noise I get from the laptop mic, I suppose because it's fan noise from a machine several years old and so is pulsing and variable in both pitch and volume. The result sounds like I recorded it at the bottom of an empty swimming pool while being attacked by some sort of Doctor Who monster ghost, and removes a lot of the bottom end of my speech range as well, which is an arse as Jeeves ended up sounding more like Sandy Toksvig.
If I don't process it, the fan noise jumps about noticeably whenever I do an edit, and work got me into the habit of doing a *lot* of edits. At times we'd even do a cut in the middle of a word because we liked the start of one take and the end of another, and we could make even mid-word transitions undetectable. With the variable fan noise, even cuts between scenes are distracting, and I find throw you out of the story somewhat. Also, because I'm used to being able to fiddle with levels on the recording to get a more consistently audible volume across things said in different ways, that just exacerbates the issue with the fan noise. I'm also used to recording two or three takes of a scene straight through, picking the best one and then doing extra takes for any lines or words I don't like and pasting them in. With what I have, you just can't do that.

So! The questions:
What do you use to record with? Work's setup was either DAT or straight into the Mac, using mostly optical cables from a soundproof booth with an expensive mic and popper stopper. This is obviously not an option, and I don't really want to spend a lot as I doubt I'll be doing a lot of this - I can only do it when everyone else is out of the house for one thing. I would like to get at least this one done, though, and I've been thinking that if I can get it done to my satisfaction and people like it I might try writing something intended to be straight-to-audio, more along the lines of a radio play with sound effects and such. I'm probably looking at a USB mic to use on the mac as that's the quietest machine and has the software for editing, are there any particularly good ones? I had a logitech stick mic which was fairly decent but which got broken, but the headset which came with my housemate's teach-yourself-spanish course had awful frequency response and crackled.

How do you approach it? Do you do the whole thing in one take and do over if it's not up to scratch, do a scene at a time, or edit things together even mid-scene? My fic comes out at about 45 minutes long so doing it in one take isn't really an option for me. I have a slight speech impediment which I can mostly control if I concentrate, but it often means that I'll choke on a phrase if it's not something I'd usually say myself, or mispronounce something and only notice it later. Even doing complete scenes at a time is going to mean a lot of retakes. Hell, it took me about 8 hours including editing time to do the unacceptable version, and that was mostly just replacing individual lines I'd messed up.
If you do split it up, do you take breaks in between, or do it all in one sitting? Do you get one scene perfect before moving on to the next, or do you go over it all at the end and polish the bits you're not happy with? The first way would help to keep the levels constant within a scene and eliminate some of the fiddling afterwards, but I suspect it would take a lot longer to actually do and might make the thing more variable start-to-finish.

Voices - Do you attempt to make the characters sound how they do in your head as you're reading/writing, or do you just try to make them different enough that it's clear which one's which when they're speaking? I find Jeeves a relatively easy character to do, because a measured middle-class RP accent is what I use all day at work (much to the amusement of my colleagues who just use their own accents and accuse me of wanting to read the news on radio 4) and Bingo I don't mind being something of a caricature but I'm not happy with my Bertie. When I tried making him sound more upper-class my pitch went up and I couldn't fix that in software because of the fan noise, so to me it sounded more like I was trying to do Honoria. I settled for doing middle-class and just leaving it slightly more animated so hopefully the personality came through enough to differentiate them.

Lastly, am I overthinking it all? Should I just get hold of a better mic, stop worrying about it, re-record and however it comes out is good enough?