-
Notifications
You must be signed in to change notification settings - Fork 12
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Issue #281 Start work on generic recognizers
- Loading branch information
Showing
3 changed files
with
540 additions
and
0 deletions.
There are no files selected for viewing
77 changes: 77 additions & 0 deletions
77
src/AnalysisConfigFiles/RecognizerConfigFiles/Ecosounds.GenericRecognizers.yml
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,77 @@ | ||
--- | ||
SpeciesName: Test Species. | ||
CommonName: Generic Test | ||
AbbreviatedSpeciesName: TestSp | ||
|
||
# Resample rate must be 2 X the desired Nyquist | ||
ResampleRate: 22050 | ||
# SegmentDuration: units=seconds; | ||
SegmentDuration: 60 | ||
# SegmentOverlap: units=seconds; | ||
SegmentOverlap: 0 | ||
|
||
#Proposed Approach: | ||
#1. Assume that a long recording has been broken into one-minute segments. | ||
#2. Convert each one-minute segment to a spectrogram. | ||
#3. Obtain a noise profile for each segment. This is to be used later to remove insect chorusing. | ||
#4. Scan the one-minute waveform and select "spike maxima" whose amplitude exceeds a decibel threshold, D. | ||
#5. Extract a single frame (say 512 samples) centred on each spike and convert to a spike spectrum. | ||
#6. Subtract the noise profile from the spike spectrum. | ||
#7. Smooth the remaining spectrum. | ||
#8. Look for evenly spaced harmonics in the smoothed spectrum. | ||
# Typically the lowest harmonic will lie between 1200 Hz and 3000 Hz and the higher ones evenly spaced. | ||
# This is the tricky bit due to variability but may work to use spectrum auto-correlation. | ||
|
||
# Each of these profiles will be analyzed | ||
Profiles: | ||
#Standard: &STANDARD | ||
#EventThreshold: 0.2 | ||
# Common settings | ||
#BgNoiseThreshold: 3.0 | ||
Territorial: | ||
# min and max of the freq band to search | ||
# MinHz: 800 | ||
# MaxHz: 8000 | ||
# MinDuration: 0.15 | ||
# MaxDuration: 0.8 | ||
# DecibelThreshold: 9.0 | ||
Wingbeats: | ||
# MinHz: 200 | ||
# MaxHz: 2000 | ||
# DecibelThreshold: 6.0 | ||
# duration of DCT in seconds | ||
# DctDuration: 0.5 | ||
# minimum acceptable value of a DCT coefficient | ||
# DctThreshold: 0.5 | ||
# ignore oscillation rates below the min & above the max threshold | ||
# A sample of four wingbeats yielded an average of 5.1 +/-0.5 beats/s. | ||
# 95% confidence interval (2 SDs) is therefore approx 4-6 beats/s. | ||
# OSCILLATIONS PER SECOND | ||
# MinOscilFreq: 4 | ||
# MaxOscilFreq: 6 | ||
# Minimum and maximum duration for the length of a true call. | ||
# MinDuration: 1.0 | ||
# MaxDuration: 10.0 | ||
# Event threshold - use this to determine FP / FN trade-off for events. | ||
# EventThreshold: 0.5 | ||
#Agonist: | ||
# This notation means the Groote profile has all of the settings that the Standard profile has, | ||
# however, the MinHz and MaxHz properties have been overridden. | ||
# <<: *STANDARD | ||
# DctDuration: 0.3 | ||
|
||
# Available options (case-sensitive): [False/Never | True/Always | WhenEventsDetected] | ||
SaveIntermediateWavFiles: Never | ||
SaveIntermediateCsvFiles: false | ||
# Available options (case-sensitive): [False/Never | True/Always | WhenEventsDetected] | ||
# "True" is useful when debugging but "WhenEventsDetected" is required for operational use. | ||
# SaveSonogramImages: True | ||
SaveSonogramImages: WhenEventsDetected | ||
# DisplayCsvImage is obsolete - ensure it remains set to: false | ||
DisplayCsvImage: false | ||
## End section for AnalyzeLongRecording | ||
|
||
# Other config files to reference | ||
|
||
#HighResolutionIndicesConfig: "../Towsey.Acoustic.HiResIndicesForRecognisers.yml" | ||
... |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.