The following is an excerpt - with minor editing - from an e-mail
message I recently sent to someone who was about to make some audio
recordings for submission to the audio archive. On 19 September 2000
I've modified the file to ask for a minimum sampling rate of 22,050 Hz
in place of the 8,000 Hz I previously asked for:
If you use Windows and if you don't already have a
good audio package, I recommend that you download GoldWave. Once you take awhile
to learn to use it, it will save you a lot of trouble. It's
shareware, so you don't have to pay for it unless you decide you're
going to continue to use it.
I'm going to mention some things to watch out for. You may already
know about them; in fact you may know a lot more about sound recording
than I do. Please forgive me if I tell you something you already
know.
Be prepared to be repeatedly dissatisfied with what you record.
Your first recording, or your second or seventh, may have pops,
clicks, or hum, and it may be recorded at too high or too low a level.
To get an idea of how many things can go wrong, listen to some of the
recordings in the AUE Audio Archive. Some of them are barely adequate, typically containing too many
pops and too much hum or other background noise.
Talk across the mike rather than into it. Talking into the mike -
that is, holding the mike directly in front of your mouth - tends to
produce pops when you pronounce words starting with 'p'. Other
plosives are also exaggerated when you face the mike directly. You
may have noticed in televised recording sessions on TV that some
vocalists have a wire mesh between the microphone and their mouth.
When you pronounce one variety of 'p', you tend to spit; spit hitting
the microphone results in pops in the recording.
Don't hold the mike in your hand; put it on a stand. Holding the
mike in your hand can result in annoying rubbing noises as the mike
shifts in your hand.
Keep the mike away from fluorescent-light fixtures, computer air
vents, and any other sources of noise you might have. The room can
seem quiet to you while the mike picks up all sorts of noises you wouldn't
suspect.
If you have noise that you can't trace to a source, it may be that
the mike is picking up vibration from the surface the stand is sitting
on. Try setting the stand on a surface that isn't vibrating, or try
putting a cushion of some sort - maybe a mouse pad - between the
vibrating surface and the mike stand.
Choose: sampling rate 22,050 Hz, 'mono', 8 bit. It would be nice to be
able to use a higher sampling rate, but the size of the file is
directly proportional to sampling rate, and it's good to save as much
Web-site room as possible. The sampling rate of 22,050 Hz is a minimum,
though. It will be used to make an MP3 file at that sampling rate. For
large files, the sampling rate may be reduced for WAV files.
Watch the oscilloscope display on the GoldWave display while you
are recording. You can tell by watching if your level is too low or
too high. Try to adjust the level so that you're filling up most of
the available amplitude range without clipping. Clipping appears as
extended flat tops on the waveform. Keeping the recording level
as high as possible without clipping gives the recorded voice an advantage
over quantization noise and other internal noise.
Adjust the recording level so that you can get the mike close to
your mouth without overloading. That gives your voice an advantage
over ambient noise.
Listen to the recording after you've made it. If you hear anything
you're not pleased with, try to figure out what went wrong. Take
appropriate corrective measures and record again. See if you can make
the quality of your recording better than any of the ones in the AUE
Audio Archive.
If something about the facilities you have won't let you record at 22.05
kHz, go ahead and record at a higher sampling frequency. The webmaster can use
GoldWave to resample the files after he gets them.
If you have any questions, or if you have comments about my suggestions, please
feel free to send e-mail to me.