Biomedical Engineering Reference
In-Depth Information
Recorded syllables
0
T(s)
Sy n thetic sy lla ble s
T(s)
0
Fig. 5.10. ( A ) a sonogram of a recorded fragment of song, and ( B ) a sonogram of
a synthetic song, generated when the model was driven by linear functions of the
smoothed envelopes of the activities recorded during vocalizations
computed with data from one syllable) are also used for other syllables. In
Fig. 5.10, we display sonograms of an actual song and one computed from
EMG and pressure data.
The measurement of the activity of the dorsal muscle dS allows us to
test the prediction described in the previous section regarding the relative
sizes of the activities at the beginning and end of each syllable, depending
on the temporal evolution of its fundamental frequency. The analysis of the
model predicts that for upsweep syllables, the minimum value of f 0 needed
to prevent oscillations at the begining of the syllable is smaller than the value
needed to stop the oscillations at the end. The ratios between the maximum
activities at the end and at the beginning of 17 syllables are displayed in
Fig. 5.11. The first six syllables are upsweeps, while the rest are downsweeps.
The latter show ratios smaller than one, as predicted.
The physical processes involved in birdsong are very rich, and yet it is
possible to attempt to produce simple models leading to qualitative agree-
ment with data. Since the origin of the sound in many examples of birdsong
has been traced to the vibration of labia, the existence of simple oscillatory
models reproducing the sounds is only natural. However, it is not trivial even
in principle to interpret model parameters in terms of the actual biological
Search WWH ::




Custom Search