orcasound / orcadata Goto Github PK

I listened the audios in TrainDataLatest_PodCastAllRounds_123567910.tar.gz (wav subfolder) and visualized the annotations and realized that they do not match. Take for instance 60012.wav and the first annotations
podcast_round1 60012.wav 34.126 2.918 Dabob Bay, Seattle, Washington 1960-10-28 60012
podcast_round1 60012.wav 36.816 2.588 Dabob Bay, Seattle, Washington 1960-10-28 60012
podcast_round1 60012.wav 42.55 2.055 Dabob Bay, Seattle, Washington 1960-10-28 60012
podcast_round1 60012.wav 44.606 2.41 Dabob Bay, Seattle, Washington 1960-10-28 60012
podcast_round1 60012.wav 46.636 3.425 Dabob Bay, Seattle, Washington 1960-10-28 60012
podcast_round1 60012.wav 51.381 3.248 Dabob Bay, Seattle, Washington 1960-10-28 60012
you will see that they onsets and offsets do not match exactly the start of the vocalizations and there are vocalizations also outside these time intervals. it looks a bit random tbh

Running WAV_FLACprocessor_Val/AnnotateSpectrograms/WAV_FLAC_Annotator_0_6.py gives the following error-

NameError: name 'sf' is not defined

The following path mentioned in the ML-resources wiki page is wrong.
https://github.com/orcasound/orcadata/wiki/Orca-ML-resources-(train-test-data-&-model-efforts)
Intermediate signal:noise ratio

05 Jul 2019 -- 1/2 hour of data from the Orcasound Lab node labeled in Audacity by Scott
Link for labels-https://acoustic-sandbox.s3-us-west-2.amazonaws.com/labeled-data/classification/killer-whales/southern-residents/20190705/orcasound-lab/test-only/OS_7_05_2019_08_24_00_labels.txt

Now there are 3 different labels files stored at this location-
https://acoustic-sandbox.s3-us-west-2.amazonaws.com/labeled-data/classification/killer-whales/southern-residents/20190705/orcasound-lab/test-only
1-OS_7_05_2019_08_24_00_labels-SV_200210_only_calls.txt
2-OS_7_05_2019_08_24_00_labels-SV_200210_other_signals.txt
3-OS_7_05_2019_08_24_00_labels-old-typos.txt
Please update the wiki page.

The Readme file says
With mmsi number go out on web and find metadata for this ship (ship name, ship type,.....)Program: parsePage.py(missing data from for the ship)
but it is missing from the folder.

AISprocessing/AWS_FlowDiagram is not a valid jpg image.

Are there some missing annotations in Haro Humpback, usually high frequency calls?
In OS_10_28_2021_18_54_00 from around 10:16 to 10:18 -> is that an orca call and this is why it was not annotated?

In file orcadata/WAVprocessor_Val/ProcessWAVsForCalls/scanWAVsForCalls.py following line is written import searchForCalls as sff
but searcForCalls module is missing.

I find the purpose of this repo kind of confusing. The name and the access.md file are focused on orca data. However, the description focuses and other folder focuses on software. The main focus seems on preprocessing software, since a lot of the more advanced software now is in other repos, so the scope can be updated.

I also find the diagram in the access.md very hard to read since it points leftwards! It most probably needs a caption, and there may be a better place for it, unless one has a better way to relate it to the access topic.

Maybe with some community suggestions/discussion the organization of this repo can be improved.

In file Spectogram/Scripts/spectrogram.ipynb, cell 2, line-

plotPath = join(basePath, "orcadata", "spectrogram", "Plots")

does not correspond to an actual file path.

Several Spectrogram Plots PNG corresponding to their sound flac data is missing. Also some Spectrogram Plots path are not correctly identified using the .split(sep="-")[1] method.

i'm trying to download a file from the aws with this:

aws --no-sign-request s3 cp s3://archive-orcasound-net/rpi_orcasound_lab/flac/2020-12-12_03-59-38_rpi_orcasound_lab--2.flac ~/orcasound.flac --storage-class STANDARD --recursive --force-glacier-transfer

I get no error, but file isn't being downloaded. maybe because it's in Glacier storage?

how can I access the archives?

spectrogram.ipynb uses soundfile library and makes a spectrogram a bit hazy,
tried doing with Librosa amplitude_to_db spectrogram
AtoD = librosa.amplitude_to_db(np.abs(librosa.stft(data)), ref=np.max)

https://colab.research.google.com/drive/1LRdUcsKTiBI-ERVJQ0OEImcUo7WqtjrN

Currently the Orcasound Creative Commons license is only described in the footer of the Orcasound Wordpress-based web site. We could also clarify what attribution should look like here (and there).