Except for the little "glitch" problem -- this is not "ess" sibilance, pops, peaks, or other audio problems, but artifacts of data loss in the recording process.
I experimented with changing device buffer sizes. The default was 512. Going up to 4096 samples was a total disaster, but going back down to 128 did not get rid of the problem. It is some kind of problem with the streaming audio that results in buffers lost and repeated. This is not too noticeable in a section of silence, of course, but if I'm in the middle of a word and a previous buffer gets "pasted" into the stream of data it will cause a sudden spike as pieces of the waveform no longer connect smoothly -- essentially, D/A converter is asking the speaker or headphone driver to accelerate in a way that isn't physically possible. So you hear noise instead. Here's what the destroyed waveform looks like; the image shows a fraction of a second of 16 bit samples, 44.1 KHz, mono:
It isn't the Snowball microphone -- the PC records its output flawlessly. The PC is no good for recording, though, because I can't get it far enough away and its fan noise comes through terribly loudly on the recording. I'd need a sound-insulated vented case or some other arrangement for it, or I'd need to record in a different room from the PC, and I just don't have a good way to do that at home; no sound booth.
I made a complete recording of the Captain Gault story number two, "The Diamond Spy." It's a great story and the recording came out very well, with the exception of a few dozen of these bursts of noise. I'd almost be willing to ignore those, but in addition, in the 30-minute recording session I got two much longer sections that have constant noise, where a long section of a paragraph or two or three contains these glitches over and over, lasting up to a minute.
That means retakes, and more editing, and edits which don't sound as good, since the background sound has changed. This quickly turns something that takes a half-hour to record raw and another half hour to edit into a two- or three-hour project. I have enough problems waiting for cars to go by, waiting for churchbells to stop ringing, etc., than worrying about an audio recording which is randomly glitched at any time without warning. It makes me want to go back to an analog recording onto something with predictable characteristics, like casette tape.
Changing USB cables doesn't fix the problem. Using a different external FireWire hard drive (and different FireWire cable) doesn't fix the problem.
However, it does seem to be related to using both FireWire and USB. If I record only to the internal drive, I seem to get a glitch-free signal.
Very strange, but it gives me a possible workaround, although one that involves yet more tedium and manual moving of files.
Maybe I just need to stick in a bigger new internal hard drive, although it seems like a waste to put money into this laptop, which has a dim screen, is on its second keyboard, and won't run off of battery power or even retain the date and time!
I really need a new machine... but I don't want an Mactel machine; they are still being shaken out, and the "thermal grease" problem in the MacBook Pro machines has made me skittish about buying one (I've assembled computers before -- that is NOT how you use thermal grease!)
I just dropped $150 on a brand-new license for the PowerPC version of a program called DSP Quattro X, which is a very nice editing program that does common operations like saving and loading files much faster than Audacity or Sound Studio 3. And despite how much I gripe about the problems I've had with my PowerBook G4/400, which is now about seven years old, the original TiBook machines are the most attractively designed laptops I've ever seen or used, period. I don't want to give mine up!
No comments:
Post a Comment