Faulty axes for FCS when using NoGaps rendering mode in ConcatenateIndexFiles

[ThomasLuypaert]

Describe the bug

When using ConcatenateIndexFiles with NoGaps rendering mode to create FCSs for long-duration intermittent recordings (1-min every 15-min), the x- and y-axes are faulty.

1. The y-axis suffers from a spacing issue (as the frequencies go up to 192 kHz, values overlap). An option to increase the y-axis tick interval would be useful.

2. The x-axis displays the wrong time. This is likely the result of the non-continuous nature of my data.

Expected behavior

A FCS with correct x- and y-axes.

Diagnostics

• [x] I have followed the instructions for reporting a bug

• I have attached a log:

  log_20190724T125240Z.txt

• The Operating system I'm using: Windows 10 Home Version 1803

• The version of AnalysisPrograms.exe: Version v19.5.0.1

• The data used can be found: Data consists 672 1-min files, representing 1-min recording every 15-min for 24h/day for 1 week, at a sampling rate of 384 kHz. A framelength of 2048 was used, and acoustic indices were calculated over 15 seconds.

• The command I used:

  D:/User/Software/AP/AnalysisPrograms.exe ConcatenateIndexFiles E:/Acoustic_Data/AM1_AIs/BatchIndicesOutput1/Week_1 -o E:/Acoustic_Data/AM1_AIs/BatchIndicesOutput1/Week_1/Output_Concatenate -d *.WAV* -ip D:/User/Software/AP/ConfigFiles/IndexPropertiesConfig.yml -fcs D:/User/Software/AP/ConfigFiles/SpectrogramFalseColourConfig.yml -f Acoustic_Test -z 0 -g NoGaps --draw-images -n --concatenate-everything --log-level 7

• The config file I used:

Screenshots

Additional Details

Add any other context about the problem here.

[atruskie]

Great catch!

For the y-axis, the current config file supports changing the YAxisTicInterval. If you bump that to 10000 it should be much better.

https://github.com/QutEcoacoustics/audio-analysis/blob/79a2f0079fd6388eab51bee433e2ae6ea2444030/src/AnalysisConfigFiles/SpectrogramFalseColourConfig.yml#L22-L27

I'm going to suggest adding an Automatic mode to the Y-Axis though that will choose appropriate values... well automatically.

---

For the x-axis, yeah you're absolutely right. For non-continuous data (i.e. the NoGaps option) we weren't sure if anyone would even use that kind of visualization. So we didn't bother fixing the axis. We are also not sure what the axis should look like. Do you have any suggestions?

Here is my proposal:

The red-dotted lines indicate non-contiguous data. Your data is regular so this would work well. For irregular data, or a mix of contiguous or not, it would need to be different.

Scenario: one-minute every 15, except for dawn (05:00), which is recorded continuously for 3 hours in 30-minute contiguous recordings (I'm using the same image, so the FCS data is wrong, but let's imagine).

I've also added a source indicator (Green-Blue alternating strip) to that image which denotes when a different audio file is used as the source data.

We should probably fix the dotted-grid markers too.

Any opinions: @towsey, @elizianeoliveira, @KellieVella, @ninascarpelli?

[ThomasLuypaert]

Your proposal to fix the x-axis looks great. Perhaps a 12-hour interval for the time ticks would be useful, so we can always see midnight and noon for each day?