- Ronald Kessel added an answer:How do you calculate the number of true negatives using an automated acoustic classifier?
I’m working on an automated frog call detection problem. I’m trying to determine a good way to estimate the number of true negatives within a given sample. The true and false positives, and the false negative are easy, but to perform some of the more sophisticated analyses, one needs a value for the true negatives. True negatives in this context can be defined as the sample space within which the automated classifier could have made an incorrect classification, but did not.
My thought was to sum the time taken up by the true and false positives then add the time taken by the false negatives (determined by multiplying the number of false negatives by the mean time of a true positive). Then subtract that value from the overall time, leaving the time that was “at risk” of incorrect classification. Then divide the remaining time by the mean time of a false negative (with the logic that if it did make a hit, it would have been incorrect) to get the number of true negatives.
The problem with this method is that it does not work when there are lots of calls, frequently the summed time exceeds the recording time, which makes sense. I am going to explore the use of the median, but I was curious if anyone else has gone down this particular path with any degree of success?
The question is interesting. I have been working on automatic detection many years, but I have never come across the need to assess "true false negatives" as you intend.
The explanation of your suggested approach is not entirely unclear to me, but it seems to be mistaken in several respects. I think you can find your answer another way.
If you have a frog-call detector that somehow discriminates a frog call from other noise (from bird calls or traffic noise and what not), or that discriminates a particular species of frog from other species, then that detector presumably has a detection-sensitivity adjustment. The adjustment probably amounts to the adjustment of one or more decision thresholds in your detection algorithm.
The only detectors that do not have such detection-sensitivity adjustments are expert systems, such as neural networks for instance, that are trained for detection in a once-and-for-all fashion, using validated populations target samples (frog calls) and non-target samples (ambient, spurious noise). The sensitivity setting for such detectors is for the most part implicitly “learned” from the relative sizes of the training populations, rather than adjusted at the time of use in the field. I always advise very strongly against using such detectors in practice, precisely because they have no field-adjustable sensitivity setting. In practice, the only time that you can get away without field-adjustable detection-sensitivity -- hence the only time that such expert systems might be of operational use as detectors -- is when the real-world field conditions are thoroughly under your control and they never change. That is rarely ever the case in reality. It is probably not the case for frog-call detectors in wetlands. So be sure that your frog-call detector is not a neural network or similar pre-trained expert system, and that it has a detection-sensitivity setting. If your detector is without a sensitivity adjustment, send me a note, and I we can discuss it further.
To answer your question then: Assume that you record audio in the vicinity of wetlands, and that you can replay it for your detector as many times as you like back in the lab. For one play of the recording you use a relatively low detection-sensitivity setting S1, and your detector counts N1 false positives in time T (i.e., N1 is the number of events that your detector identifies as frog-calls when they are not in fact frog-calls).
Now increase the detector sensitivity somewhat to S2, such that the expectation for false positives increases. Replay or continue to play the recording for the same time length T, and register the number of false positives N2. You should have N2>N1 because the detection-sensitivity is greater for S2 than S1.
The difference N2-N1 is the number of true false positives that were correctly rejected by your detector in the special case of going from sensitivity setting S2 to S1.
Now increase the detector sensitivity once again, to S3, and record the number of false positives N3>N2>N1. Keep on going this way, systematically increasing sensitivity S in steps, recording the data (S1,N1), (S2,N2), (S3, N3)… and so on, until the number of false positives N becomes the largest number that you care to count in time T.
Plot the points (S,N) on a graph. Draw a smooth-line fit (S,N) through the points. The value of N on this line at the maximum sensitivity S that you used is the number of true false positives.
If N is getting too large to count, then you might consider decreasing the time T of audio play to T’. Record N’ for this shorter time, and plot N = T’ X N’ / T on the graph. Then you can extent the line (S,N) to greater sensitivity S than you otherwise might have.
In any case, I think you may find that the events that constitute false positives are not quite what you expected them to be. This is because the detector hears the world through the particular feature set that it was designed to use for recognizing frog-calls amid noise, which is probably different from the feature set that you as a frog-call expert use to identify frog calls. It is this that makes the class of true false positives interesting. Looking at true false positives amounts to looking at (listening to) the world through the eyes (ears) of the detector, before it has made any decisions about what it is hearing. This will generally be very different than the way humans see (hear) the world. The set of true false positives may be another way to understand what the detector is doing in practice.
I hope that makes sense and helps.
(PS: I lived many years nearby in Halifax.)Following
- Niklas Klügel added an answer:How do I reduce the background noise in a playback track without affecting the vocalization of interest?
I'm trying to reduce the background noise of an audio track to use it as a playback track, but I'm afraid to affect the sound of interest and ruin my experiment.
Is there a "right way" to do it?
as previously mentioned it depends on the characteristics of the "background noise". if you have level differences (background noise has a significantly lower loudness level), then another approach (more from the recording industry) is to use gating or upward compression/expansion. You can do either in hardware or plugins in realtime or in batch (depends on your environment).Following
- Patrícia F Monticelli added an answer:Bioacoustics and saline environment, is there any study about?
I am trying to understand signal evolution in terrestrial mammals in an environment such as neotropical beach (south of Brazil, municipality of Florianópolis,Santa Catarina), humid, saline, windy versus a rural more savana like environment. I know the classical studies of Richards and Willey, Aubin and Jouventin, but I could not find any study relating saline or humidity degrees with signal funtional structure. Any possible help? Any comparison made about bird species living in coastal versus savanna environment?
Dear Hervé Glotin.
Lets follow each other here in research gate! I am preparing papers yet, that shall discuss this. Did you mean researching together?Following
- Olesya Anatolyevna Astakhova added an answer:In clades that duet, why does one species or subclade lose their duet?
And more broadly, are there some good cases of bioacoustic signals being lost where the reason for the loss is well-studied?
The crane (screaming birds) developed duets between female and male. In passerine birds developed song duets between males.Following
- Vladimír Nemček added an answer:What do you think about acoustical monitoring of owls with commercial recorders such as Olympus DM650?
We (some Czech and Slovak ornithologists) use commercial recorders for owl monitoring (in a weatherproof box placed on the tree).
I am looking for other people who use this method or similar.
Thank you for answers. I also prefer passive monitoring before call broadcast, but also I agree that it is very effective method. To Olympus recorders - I have experiences with WS 811 and DM 450 resp. DM 650. All can be used by bioacoustical monitoring (WS with some limitation).Following
- Ingrid Boucaud added an answer:Does anyone know this kind of great tit vocalization? May be a contact call?
I’m recording great tit vocalizations around the nest during breeding. Males can use songs, alarm calls or this kind of call that I am not sure to identify (see the attached files).
Thanks, xeno-canto is an interesting website. But it seems that there is no really a word for the calls that I am looking for. That amazing how great tits are well studied and how few is known about their vocalizations.
I have already contacted Torben Dabelsteen, I am waiting for his answer. Effectively, I hope he could have an idea about these vocalizations.Following
- Alberto Behar added an answer:Decibelimeter model for bioacoustic research?
I am looking for a decibelimeter for bioacoustic research that is not as expensive as a Brüel & Kjær and that has an enough quality. Any tips?
Larson Davis SLM are of comparable quality to B&K and the price is much lower.
- Nicolas Cusseau added an answer:What is the best available probe/way to measure particle motion (velocity/acceleration) of an underwater sound?
I am looking for a way to estimate particle acceleration of an underwater sound produced by aquatic animals, per e.g. Something easy to use in the field and reliable. Like a vector sensor or an underwater geophone?
If I model the acceleration as a = v*2*pi*f (where f is the centroid frequency), then my main concern is no longer the bounds on the velocities estimated. It is the frequency estimation because I assume the diver or the fish is introducing "micro"-Doppler effect.
Have a look at a fish school on ADCP data to be convinced ;).
I know that AVS is the new trend sponsored at ECUA but could someone point us to a good scientific paper (that should be the point on ResearchGate) with directivity index, gain and bounds on the frequency-dependent DOA estimated instead of commercial articles about the Hydroflown?Following
- Marie A Roch added an answer:Which is the best software for marine soundscape analysis?I am new to the soundscape analysis field. I am looking for a software that would allow me to analyse different metrics, from the basics (amplitude of a certain frequency band) to the advanced such as counting of specific signals (snapping shrimps snaps and fish vocalizations) and diversity indexes. I have limited experience with R and no experience at all with Matlab but I am willing to invest time in it if it is really worth it.
There is no magic bullet that will let you detect every type of signal. For general data exploration, I'd suggest the following software packages: xbat (Matlab program written by Harold Figueroa from Cornell - nicely done, no longer supported and documentation is supposed to be a bit on the slim side),. Cornell's Raven, Osprey (Matlab program written by Dave Mellinger at Oregon State that has some nice annotation features), or Triton (Matlab program written primarily by Sean Wiggins at Scripps Institution of Oceanography that is designed for very large datasets and has a nice compressed spectrogram function [some of the other packages have this as well]). There are other packages out there as well. For detecting calls, it really depends upon your soundscape and the calls you are interested in as to how easy that is to do. The Teager energy detector was used by Kandia and Stylianou (2006) for detecting odontocete echolocation clicks and should work well for snapping shrimp; we developed it independently and while we never published as Kandia & Stylianou beat us to the punch our description can be seen in some our papers (look at Soldevilla et al. 2008 or Roch et al. 2011).
I haven't done work on fish calls, but I would expect that for pulsed calls you would want to key in on some aspect of the frequency range and pulse rate.
If you just want a general signal detector, you could look at the work of Erbe and King (2008), they designed a signal detector that is very easy to implement and can do a pretty nice job.
Best of luck in your research - Marie
Erbe, C., and King, A. R. (2008). "Automatic detection of marine mammals using information entropy," J. Acous. Soc. Am. 124(5), 2833-2840.
Kandia, V., and Stylianou, Y. (2006). "Detection of sperm whale clicks based on the Teager-Kaiser energy operator," Appl. Acous. 67(11-12), 1144-1163.
Roch, M. A., Klinck, H., Baumann-Pickering, S., Mellinger, D. K., Qui, S., Soldevilla, M. S., and Hildebrand, J. A. (2011). "Classification of echolocation clicks from odontocetes in the Southern California Bight," J. Acous. Soc. Am. 129(1), 467-475.
Soldevilla, M. S., Henderson, E. E., Campbell, G. S., Wiggins, S. M., Hildebrand, J. A., and Roch, M. A. (2008). "Classification of Risso's and Pacific white-sided dolphins using spectral properties of echolocation clicks," J. Acous. Soc. Am. 124(1), 609-624.Following
- Jonas Pederassi added an answer:I'm revising the Pseudopaludicola genus based on its bioacoustics. Could someone help me with recorded calls?Advertisement call of Pseudopaludicola species.
Herman, maybe you can help me! I need a statistical test to compare de calls of very similar species (to have a level of confidence that they really are different). Do you know a test like this? Thanks...Following
- Gianni Pavan added an answer:Which acoustic software do you use for simulations (underwater or outdoor propagation)?I usually deal with outdoor noise propagation (noise impact assessments). I use software like Cadna, Soundplan and Predictor. I'm Interested in bioacoustics (I don't do this for a job) I'd like to know if specific software (free and not) exists to study the noise effects on animals. Sometimes reading noise reports that deal with impacts on animals I noticed that usually a dB(A) is used. Is it correct that a human filter scale can study the noise effects on animals?
Propagation plays an important role and in the marine environment it is really complex.
The best available free software is ActUp, not easy but powerful:
- Carlos B. Araújo added an answer:Is it possible to estimate SPL (dB) of bird calls from .wav file (audio recording)?I am working on avian acoustic adaptations in urban habitats. I use Avisoft SAS LabPro and Raven Pro1.4 software. These software do give amplitude levels of sound but are exceptionally too high in comparison to measurement made by Sound Level Meter SVAN 957. What measurement setting should I do in the software for getting accurate SPL( dB) from the .wav files?
You can do that if you calibrate your recorder using a sound source of known Sound Intensity Level. Still, you have to understand a little the differences from Leq, RMS and Peak.Following
- Laura Kloepper added an answer:Is there any clue in song sonogram or spectrogram to understand which structure is producing the sound?I'm working with sound emission of Netrosoma (Orthoptera) from Mexico, together with Paolo Fontana. We are trying to understand which body parts are involved in sound emission since some spp are without stridulatory file.I agree with Charles Henry about anatomical validation. In our lab we have modified a GoPro action camera (which records at 240 frames per second and fantastic resolution) with macro lenses and have been getting incredible "high speed" video of echolocating bats for a fraction of typical high speed camera costs. Contact me directly if you'd like more information on this setup--all the components are available commercially from various companies.Following
- Scott R. Veirs added an answer:What is the best sound production software to produce sounds from scratch?What is the best software to produce sounds from scratch with sequences/pulses of different frequencies and intensities, for a playback study?I vote for (and use) opensource, free, platform-independent Audacity -- http://audacity.sourceforge.net/Following
- Brandon Peterson added an answer:What is the lowest frequency of acoustic waves able to be generated in a liquid interface?Biomedical use of ultrasonic acoustic waves usually use frequencies in the MHz or sometimes in the kHz ranges. I am working on a project that requires much lower frequencies, and would like to know the experiences of anybody using instruments currently available to produce these low frequencies.Dear all,
No worries about "mis"information. I appreciate the forum of discussion, as multiple viewpoints always tend to generate better science than formulating an experiment based on a sole idea. As a microbiologist, waves are outside my area of expertise, so I have learned a lot already from the discussion above.
Many thanks for all the responses.Following
- Jun-Xian Shen added an answer:Are there italian studies on the effects of noise on animals?I m interested on the effects of noise on animals, in particular in studies made in ItalyYes. I am still interested the studies on the effects of noise on animals, especially frogs in China. One new paper is just published on JCP A, 2014.Following
- Christina Perazio added an answer:How to determine the pulse repetition rate of dolphin burst-pulse sounds?I am trying to categorize burst-pulse sounds of dolphins within a data set, and am having trouble figuring out how to determine the pulse repetition rate of a sound. Raven (Cornell software) does not seem to have any instructions on how to do it in their software, but is this something another software can do? Can you count the number of pulses visually? Do I need a code, for example in Matlab? Any help would be greatly appreciated!Thank you very much. I agree-the sound classification based on 'sound names' gets confusing with different authors referring to similar sounds with varying names, or very different sounds with similar names. I wanted to be more precise by using sound categories, so I have used the inter pulse interval to create classes for sounds as you suggest. I am looking into the soundruler software, since it seems to be very useful, but am having difficulty getting it to work on my mac, but I will contact the company. Thank you again!Following
- Javier Almunia added an answer:Do wild dolphins experience age-related hearing loss, i.e. presbycusis, as is common in humans?Humans aren’t the only ones who lose their hearing as they grow older. Scientists report that wild Indo-Pacific humpback dolphins (Sousa chinensis), which can live 40-plus years, also have trouble picking up sounds as they age.
http://jeb.biologists.org/content/216/22/iii.fullThe National Marine Mammal Foundation has a research program on aging in dolphins (http://nmmf.org/aging/#more-209)
And they have published results on the hearing loss with age, also early onset of hearing loss on males
- Ciro Alberto Sánchez added an answer:Does anyone have a protocol on how to calibrate a portable audio recorder?I am working on underwater environmental recordings made with an hydrophone (Hi Tech HTI 96 min) and a portable recorder (sony PCM M10). I would like to convert my relative negative dB values in SPL dB re 1 uPa. So far I know that I need to feed a pure sine wave in the recorder with a signal generator, read voltage with an oscilloscope, make a recording and then put these readings together with a software like Avisoft or Raven. I have access to the instruments and the software but I find it really hard to obtain detailed info about the practicality of this process. Does anyone have experience with this calibration and maybe have a protocol? I would be immensely grateful because this thing is driving me crazy.Tullio,
You can research on CEM, Spanish Center of Metrology web site: www.cem.esFollowing
- Gowtham Chitimireddy added an answer:Where can I find indications regarding the application of Directive 2008/56/EC in Italy?I am interested especially in the noise section.even though it sound's ridiculous these details are considered confidential. so, the only place you can find solution for this in EC office in belgiumFollowing
- J. Patrick Kelley added an answer:Is anybody using CoolEdit for the analysis of underwater field recordings?I've recorded several tracks in wav format (96KHz, 24 bit, which is the maximum "resolution" allowed by my recorder), but when I open these tracks in CoolEdit and I try to zoom in the recordings, a weird thing happens (Ive already used this program and this is the first time I' ve this problem): the spectrogram is grainy...seems like a picture taken at a very low resolution in which you can detect pixels. Does anybody know how to solve this problem?So, did the old version of CoolEdit have a maximum allowable sample rate?Following
- Ingrid Kaatz added an answer:What is the best digital recorder to record sounds produced by bottlenose dolphins (Tursiops truncatus)?I am starting Master and I'll study about bottlenose dolphins' bioacoustic. I need to buy a new recorder and I'd like know what is the best.OOPS
learning how to attach pdfs!
Here is the bioacoustic tech paper I noted above!Following
Bioacoustics refers to the investigation of sound production, dispersion through elastic media, and reception in animals, including humans. This involves neurophysiological and anatomical basis of sound production and detection, and relation of acoustic signals to the medium they disperse through. The findings give us some evidence about the evolution of acoustic mechanisms, and from that, the evolution of animals that employ them.