Page 20 - Acoustics Today

alkaline battery. The Roland R-26 requires nearly 190 Amp- hours, and might be powered by an array of 128 Watt-hour rechargeable prismatic LiFePO4 batteries, each weighing 1.4 kg. The Centon delivers compromised recordings designed for human speech, while the R-26 delivers up to 6 channels of high definition audio.
The maximum recording time of a recorder is a function of storage capacity and recording bit rate. Many of these units use high capacity secure digital (SDHC) memory cards to store audio data. SDHC cards have a current maximum capacityof32GBinapackagethatis32mm×24mm×2.1 mm. The SCXC standard, which shares the same physical dimensions, has a maximum capacity of 2 TB, and 128 GB cards are currently available. Two channels of CD quality audio (44.1 kHz sample rate, 16 bit data) will fill up a 32 GB card in about 50 hours. Compressed audio storage signifi- cantly extends recording time, and also reduces power con- sumption because of reduced energy consumption for writ- ing to memory. In a 128 kbps MP3 format, that 32 GB card will last almost 23 days.
Digital audio recorders are often equipped with micro- phones and one example is included in Table 1. In cases where external microphones are preferred, the preamplifier and recording sections can be utilized separately. Microphone preamplifier noise levels are often so low they
are difficult to measure accurately. EIN is an equivalent input noise estimate given the output impedance of a typical microphone, usually 150 ohms. A comparison between the recorder EIN and microphone self noise reveals that the recorder noise is often insignificant and the microphone will be the limiting factor.
There is a loss of information and addition of artifacts when audio compression reduces the storage requirements of a signal, though the psychoacoustic perception algorithm underlying MP3 encoding seeks to minimize the salience of these defects. Acoustic sources and animal calls are still read- ily recognized by active listeners and automatic detection and classification software should be effective with compressed audio of this quality. A recent study has shown that the ener- gy of the signal is preserved through the audio compression
16
This allows for calibrated sound pressure levels to be calculated, although the accuracy of these measurements will be less than the reference instru-
in a one-third octave band sense. ment used to develop the calibration.
Software for automated detection, characterization, and classification
The information presented above can help inform devel- opment of acoustical monitoring systems with high perform- ance/price ratios, but efficient data collection is moot without
Table 2. Specifications of a representative sample of digital audio recorders (DAR)
Bioacoustical Monitoring in Terrestrial Environments 19