Re: inexpensive recording technology

[mup1dm <mup1dm@xxxxxxxxxxxx>]

There are several facets to consider in making a recording for subsequent
analysis.

One aspect is the sound level recorded onto the tape. The voice needs to
be loud enough that it overcomes the inherent noise of the tape (or
whatever) but not so high that it distorts. In practice, this generally
requires someone setting a knob. Inexpensive recorders have "automatic
level control" that will reduce the level of loud sounds and raise the
level of soft sounds to keep everything within the acceptable range of
the recorder, but these tend to give recordings that have background
noise intruding wherever the voice level is soft.

Cassette recorders such as the Marantz PMD series portables and the Sony
"pro walkman" recorders (the latter generally available from J&R Music
World in NYC) are good quality
cassette recorders and also have level controls. The cheapest Marantz
starts at about $240, however, and it can be difficult to find where they
are sold. (Try the pro-audio supplier
Full Compass, at (800) 356 5844) (Madison Wisconsin)

Another aspect is extraneous noise. The recorder itself makes noise, so a
mic attached to the recorder is usually not a good idea as the sound will
carry through the plastic to the mic directly. So a recorder with a jack
for an external mic is best. This also allows the recordist to get the
mic close to the mouth of the person talking, which helps to get a much
better ratio of voice to everything else. Noise from mic handling can
also be a problem, so either use a mic that is securely fastened to the
talker (headworn or lavelier) or a handheld mic that has "professional"
caharacteristics -- purchase it from a music store or pro-audio supplier
and have them make a cable to adapt it to your recorder.

This brings up the fact that some mics have a funny characteristic that
the tonal balance changes with distance from the sound source -- you get
more bass as the mic gets closer to the lips. If the ratio of fo level to
other frequencies is important in the analysis, then you should stick to
an "omnidirectional" mic, which will not have this characteristic, rather
than an "unidirectional", "shotgun", "cardioid" or other "pressure
gradient" mic that will.

Cassettes are inherently unstable in pitch. If you record a pure tone of
fixed frequency on a cassette it will play back with warble (frequency
modulation). If you play it back on a different player, you are unlikely
to get the same frequency at all. A digital recording will be easier to
analyse.

DAT recorders (digital audio tape) are probably the best portable digital
recorders, but starting price is in the range of $600. Mini-disk
recorders are much cheaper, but may be a very poor choice for subsequent
analysis, because in a physical sense the recording is very noisy -- they
sound good because the noise is placed in time and frequency in such a
way that it is masked by the human auditory system. While the noise may
not be particularly audible to us, it might be a problem for analysis
equipent.

If a researcher will be carrying a laptop computer anyway, then the best
quality answer for the least money might be to use any inherent sound
capability on the computer to make the recording. Save disk space by
recording monaural sound files at a low (e.g.: 22000hz 22 kHz) sampling
rate, *without* compressing to MP3 or whatever. Save more space by using
PKZIP or Stuffit to reduce the file size after the fact. A headworn
computer mic will always give decent quality sound if the input levels
are set correctly.

If pitch and automatic level control are not going to cause a problem in
the analysis, then a good ald-fashioned $30 mono cassette recorder of the
type we used to use in classrooms in the '70's will be just fine if a
good microphone is used.

Douglas McKinnie
University of Surrey