Seismographical data have been recorded by a digital seismograph at a sampling frequency of about 100 Hz (100.1603 Hz). The total amount of processed data were 44 files, each of them taking almost 990 Kbyte and containing a 27 minutes recording of the Mt. Etna volcano activity (so to have a huge recording of about 1188 minutes, i.e. 20 hours). Here follows an example taken from one of the
ASCII files processed:
Starting time: 15/06/2001 00:03:39.920
Frequency: 100.1603 Hz
Samples: 168960
44
43
42
44
44
46
43
45
…
After a first scaling procedure (to properly arrange the samples in the [-1,1] interval, according to their sampling frequency and the desired resample factor (the user can specify in the program a certain resample frequency with a decrease or increase the pitch, with a remarkable effect in the auditory discrimination), an array of audio samples is constructed.
The waveform coded in the audio file will have exactly the same regularities, also recognizable thanks to the presence of some higher lines in the spectrum file.
The order of magnitude of the frequency of quasi-regular phenomena is in the range 0-50 Hz, with a spectral envelope centered around 25-30 Hz.
Setting resample factors in the Java code greater than 1 (up to several dozens or several hundreds) won't preserve the original pitch, allowing a frequency shift which will make audible regular phenomena happening at very low frequencies. In this way it is possible to observe and study (as we shall show in the next session) periodical patterns, regular behaviors, long-range correlations, which can happen at different time scales.
Starting from the scaled and resampled collection of values, to generate the waveform to store on disk or to play, it have been implemented two interpolation methods: linear interpolation and sample-and-hold (SH) interpolation. Even if linear interpolation returns particularly "smoother" sound strings, nevertheless, for some analysis application, it could be useful to prefer the SH audio file, where no external values are added to the original (scaled) data set.
