Dan, what is "acceptable" changes over time, depending on what can be
reasonably expected in terms of methodology. Now it is so easy and cheap to
record in uncompressed formats that for most situations there is no good reason
to accept the possibility of compression artifacts. Of course, that doesn't
exclude the use of compressed audio in science completely, but as general rule
I would advise people to avoid it, unless you really know your way around
compression algorithms and can demonstrate that your measurements are not
affected. But even then: recordings are often also used for purposes other than
just spectral analysis. Playback experiments, for example, where previously
recorded sounds are presented to animals to test for a response. What if Laura
or someone else later wants to use her recordings for this? Mp3 and atrac are
based on characteristics of human auditory perception, but who knows what a
frog or bird hears? We cannot assume it is the same as a human (in fact we know
it isn't), so lossy compression algorithms should really be avoided.
Best, Gabriel
it is so easy and cheap to record in uncompressed format
--- In Dan Dugan <> wrote:
>
> > Well, at least one other thing that went wrong is that it is recorded in
> > mp3. Others have said this, but it may be good to reiterate that for
> > research purposes this is simply unacceptable and will lead to problems
> > with peer review later.
>
> I suppose so, because of the hide-bound mentality of academics. But in fact,
> higher-bit-rate ATRAC- or MP3-encoded audio is every bit as good as
> uncompressed WAV files for spectral analysis. Over the many years that I
> recorded with MiniDisc (ATRAC compression), only once did I see an artifact
> in a spectrogram attributable to the compression system, and that was obvious
> and didn't interfere with the interpretation of the graphic. Walter Knapp's
> ATRAC-encoded frog recordings were accepted happily by scientists.
>
> Of course low-rate data compression is crap; but I don't think one should
> devalue a useful process because it can be misused. Nowadays memory and
> storage are much cheaper, and the only situation where one might want to use
> data compression might be very long monitoring, like 24 hour runs.
>
> -Dan
>
|