Spectrograms and Oscillograms: This is an oscillogram and spectrogram of the boatwhistle call of the toadfish Sanopus astrifer . The upper blue-colored plot is an oscillogram presenting the waveform and amplitude of the sound over time, X-axis is Time (sec) and the Y-axis is Amplitude. The lower figure is a plot of the sounds’ frequency over time, X-axis is Time (sec) and the Y-axis is Frequency (kHz). The amount of energy present in each frequency is represented by the intensity of the color. The brighter the color, the more energy is present in the sound at that frequency. 

Context in source publication

Context 1

... travels 4.5 times faster in seawater than in air (Giancoli, 2000) and can convey information in situations in which visual or chemical cues are ineffective, such as in darkness or in turbid water. Many aquatic vertebrates utilize acoustic signals for communication. Fishes produce sounds for many reasons including feeding, aggression, competition, and reproduction. The study of fish sounds is important because field monitoring can provide information about the species identity, geographic range, health, and behavior of calling individuals. These data can be collected without visual confirmation or observation if species-specific sounds and associated behaviors are known. Fish sounds are characterized by the acoustic parameters of frequency, amplitude, and temporal (timing) measurements. These measures can be used to define and compare sounds of different species and individuals. Hydrophones have been the standard equipment used to take scientific recordings of underwater sounds. These systems have historically been very expensive and custom- made. In contrast, compact point-and-shoot digital cameras are affordable and becoming more advanced. If expensive hydrophone systems could be replaced with consumer electronics without a loss of quality and necessary detail in underwater recordings, the cost of underwater acoustic research could be greatly reduced and the amount of underwater acoustic data substantially increased. The goal of this study was to evaluate the capability of point-and-shoot digital cameras to record underwater sound, and to determine if they are accurate enough to be used as tools for the scientific study of fish sounds. Simultaneous recordings were made of controlled sound playbacks to test the capabilities of the cameras. Four digital cameras (Canon Powershot A570IS, Canon Powershot SD900, Sony Cybershot DSC-P150 and Olympus Stylus 1030SW), a hydrophone system, and a recreational underwater microphone system were used for comparison (Figure 1). All of these recording systems had automatic gain control and adjusted the sensitivity of the microphone based upon the amplitude of the sound. The volume of the experimental tank was approximately 1892 liters and filled with salt water (35 ppt, 19.17 ° C). Water depth was 103 cm and the underwater speaker was placed mid-depth at 48 cm. A Clark Synthesis (229F) underwater speaker (response range of 5-200 kHz) and amplifier were used to play the sounds. Cameras were held underwater with microphones facing the speaker at a depth of 15 cm and a distance of 1.8 m from the speaker. Test sounds were created using a Digital Function Generator from Digital Recordings ( products.html) and varied in frequency, amplitude and temporal parameters. Frequency accuracy was tested using pure tones of 0.5 s duration (at 100, 300, 500, 1000, 5000, 6000, 7000 and 10,000 Hz) and a 5 s frequency sweep ranging from 100-10,000 Hz. Spectrograms from the recordings of each system were compared to the known frequency of the source signal to determine the accuracy of frequency representation for each system. Spectrograms and selection spectra were also examined to determine upper detectable frequency limits for each recording system. Amplitude accuracy was evaluated using a 5 s amplitude sweep ranging from -80 to 0 relative decibels at a frequency of 300 Hz. Oscillograms of the simultaneous recordings from each system were compared to determine the relative recording amplitude for each system. Referenced sound pressure levels of test sounds were not measured in this study. Temporal accuracy was determined using four consecutive pulses of 0.1, 0.3 and 0.5 s length at a frequency of 300 Hz, each separated by one second of silence. Pulse length measurements were made from oscillograms, including and excluding reverberation, and then compared to the known source signal length to determine the temporal accuracy of each system. Oscillograms and spectrograms of the recordings for each trial were analyzed using Raven v. 1.2.1 (Cornell Lab of Ornithology). An explanation of spectrograms and oscillograms can be found in Figure 2. The Olympus Stylus 1030SW camera is waterproof without a housing to a depth of 10 m. Additionally, a housing is available (Olympus PT-043) that allows the camera to be used down to a depth of 40 m. We were able to determine the effect of the housing on recordings by using two of these cameras simultaneously. The Olympus 1030SW was included in all recording trials both within and without a housing. A second recording trial was conducted with the Olympus 1030SW without a housing and within a flooded housing to determine if effects were caused by the air contained within the housing, or the housing materials themselves. Lack of redundant equipment prevented the simultaneous test of all three housing treatments (without a housing, within a housing, and within a flooded housing). Test sounds in this trial consisted of a 0.5 second 1,000 Hz tone and an amplitude sweep of -80 to 0 relative decibels at 300 Hz. Oscillograms and spectrograms were compared to identify effects of the housing on frequency, amplitude and temporal timing of the recordings. Many of the systems produced inherent electronic noise that infringed on all recordings (Figure 3). The Canon A570IS was the noisiest, producing the highest amplitude noise at 3 kHz, and lower intensity noise at 100 Hz intervals from 0-3 kHz. The Sony PC-120 in the Mako housing was the next noisiest system, but produced mainly high frequency noise at 8, 16 and 16.5 kHz. The Canon SD900 produced low frequency noise (300 Hz) and the Sony Cybershot DSC-P-150 produced high frequency noise (7.6 kHz) at similar levels, but both lower than the above mentioned systems. The hydrophone and the Olympus 1030SW were the quietest recording systems and did not appear to produce inherent noise. The hydrophone, underwater microphone, Olympus 1030SW without a housing, and the Sony Cybershot DSC-P150 did detect considerable noise in lower frequencies. However, this noise was due to ambient sound from aquarium pumps and filters present in the experimental room. This was verified by completing 10 second recordings of ambient noise in a different room using the Olympus 1030SW without a housing, the Sony PC-120 in the Mako housing, the Sony Cybershot DSC-P150 within a housing (MPK-PHB), and the Hydrophone (in water). The signals present on the ambient noise recordings differed from the trial recordings, indicating that the noise was due to external sources and not from the cameras themselves. Frequency bandwidth, or the range of frequencies, detected by the cameras was lower than that of the hydrophone (Figure 4). The hydrophone system was capable of recording frequencies up to 16.5 kHz while the Olympus camera was capable of recording the lowest frequency bandwidth of 4 kHz. The Sony Cybershot DSC-P150 had the best frequency range of the digital cameras, and was able to record sounds up to 10 kHz, almost twice as high as the next best range of 5.5 kHz from the Canon cameras. Frequency trials indicated that each of the cameras could accurately record frequency parameters within their respective range. In most cases, sound above upper detectable limits was not represented in the recordings. However, the Canon cameras created a sound artifact when recording sounds of frequency higher than their 5.5 kHz limit. Frequency tones higher than the limit were represented by inaccurate tones that were lower than the source signal (Figure 5). This was likely due to a phenomenon called digital aliasing, which occurs when a frequency is more than half of the sampling rate of the recorder. In this situation, the sampling rate is insufficient to accurately describe the frequency, and causes it to appear different than the source signal. The cameras that did not show this artifact may have built-in anti-aliasing filters or a higher sampling rate to deal with this problem. Amplitude trial results revealed variation in the microphone sensitivity between systems. The hydrophone and underwater microphone were the most sensitive recorders. Of the digital cameras, the Sony Cybershot had the highest amplitude recording while the Olympus 1030SW in a housing had the lowest (Figure 6). The hydrophone and underwater microphone systems also recorded higher levels of background noise than the digital cameras. However, the cameras, being less sensitive recorders, have a higher amplitude threshold for sound detection. Sounds must be louder for the cameras to detect them; therefore these systems may not be appropriate for recording low amplitude sounds. Temporal trial results (Table 1) indicated that measurements excluding reverberation were more accurate than those including reverberation, regardless of the recording system used . Average overestimation with reverberation was 137 ms, and 52 ms without. Reverberation was often easier to distinguish in recordings from the hydrophone and underwater microphone compared to the digital cameras because the waveform was more clearly represented (Figure 7). Clarity of ...