Dan Dugan, you wrote:
> I think that's the weakest point--where a distinction in noise level
> of perhaps one or two dB is being made, trying to match clock ticks
> by ear is too chancy for me. I suggest a 1 KHz tone from a speaker
> between the clocks for level matching, calibrated by measuring
> voltage at the speaker terminals. You can leave it out of the
> published movie.
I agree that the short ticks of a clock are probably a bit too
variable for an exact gain matching of the different recorders.
However, even a 1 kHz sine signal played through a speaker in a more
or less reverberant room might cause other problems. The delayed
short-term echoes will produce interferences (superposition) that can
also lead to relatively high measurement errors.
An even better calibration method would be to bypass the acoustic path
and to use an electrical reference signal, fed directly into the
microphone input. In this way, it is also possible to determine the
absolute input sensitivity (the input clipping level or "gain") of the
recorder. This is what I did in my tests. I played a synthetic 1 kHz
sine signal through the head phone output of my (battery-operated)
Edirol R-09. I measured the rms voltage of that test signal by using a
common multimeter (435 mVrms =3D -5.03 dBu in my case). I then connected
the R-09 output with the microphone input of the recorder by using a
simple voltage divider (two resistors, 100000 and 150 Ohms, providing
an attenuation of -56,5 dB; the 150 ohms resistor connected to the XLR
pins 2 and 3). This electrically recorded test signal creates a
certain dBFS level (for instance -12dBFS) in the recorded .wav files,
which can be used to normalize the recordings across the various
recorders.
Nevertheless, I would claim that even a subjective clock tick
calibration method by ear is probably sufficient for our purposes.
Yes, it is true that there would be some uncertainty in the range of 1
or 2 dB. However, such small noise floor differences should be
irrelevant when using the equipment under real-world conditions in the
field. For instance, if you managed to half the distance between the
microphone and the subject, you would improve the signal-to-noise
ratio by 6 dB.
Therefore, I do believe that the influence of the recorder and even
the microphone is small compared to the specific circumstances
(including personal experience, luck and patience) in the field.
Regards,
Raimund
> Rob Danielson wrote:
>
> >Please do point-out where the method method is lacking...
> >
> >The only ear evaluation _I_ do is match playback levels. I make the
> >high res files available so folks can tweak a test segment up or down
> >a dB or two to match to their tastes which is quite easy and quick to
> >do with QT Pro I might add.
>
|