# Access the arguments of the threshold_detection function
?threshold_detection

# Set each argument according to the targeted audio events
TD <- threshold_detection(
  CATBIC, # Either a path to an audio file (see ?read_audio), or a Wave object
  threshold = 12, # 12 dB SNR sensitivity for the detection algorithm
  time_exp = 1, # Time expansion factor of 1. Only needed for bat recordings.
  min_dur = 140, # Minimum duration threshold of 140 milliseconds (ms)
  max_dur = 440, # Maximum duration threshold of 440 ms
  min_TBE = 10, # Minimum time window between two audio events of 10 milliseconds
  max_TBE = 5000, # Maximum time window between two audio events, here 5 seconds
  EDG = 0.996, # Temporal masking with Exponential Decay Gain from 0 to 1
  LPF = 10000, # Low-Pass Filter of 10 kHz
  HPF = 1000, # High-Pass Filter of 1 kHz
  FFT_size = 256, # Size of the Fast Fourrier Transform (FFT) window
  FFT_overlap = 0.875, # Percentage of overlap between two FFT windows
  
  start_thr = 25, # 25 dB threshold at the start of the audio event
  end_thr = 30, # 30 dB threshold at the end of the audio event
  SNR_thr = 10, # 10 dB SNR threshold at which the extraction of the audio event stops
  angle_thr = 45, # 45° of angle at which the extraction of the audio event stops
  duration_thr = 440, # Noise estimation is resumed after 440 ms
  NWS = 1000, # Time window length of 1 s used for background noise estimation
  KPE = 1e-05, # Process Error parameter of the Kalman filter (for smoothing)
  KME = 1e-04, # Measurement Error parameter of the Kalman filter (for smoothing)

  settings = FALSE, #  Save on a list the above parameters set with this function
  acoustic_feat = TRUE, # Extracts the acoustic and signal quality parameters 
  metadata = FALSE, # Extracts on a list the metadata embedded with the Wave file
  spectro_dir = file.path(tempdir(), "Spectros"), # Directory where to save the spectrograms
  time_scale = 1, # Time resolution of 2 ms for spectrogram display
  ticks = TRUE # Tick marks and their intervals are drawn on the y-axis (frequencies) 
) 

# Get the number of extracted audio events
nrow(TD$data$event_data)

# Let's try various settings, starting with 1024 FFT size instead of 256.
TD <- threshold_detection(
  CATBIC, threshold = 12, time_exp = 1, min_dur = 140, max_dur = 440, 
  min_TBE = 10, max_TBE = 5000, EDG = 0.996, LPF = 10000, HPF = 1000, 
  FFT_size = 1024, FFT_overlap = 0.875, start_thr = 25, end_thr = 30, 
  SNR_thr = 10, angle_thr = 45, duration_thr = 440, NWS = 1000, 
  KPE = 1e-05, KME = 1e-04, settings = FALSE, acoustic_feat = TRUE,
  metadata = FALSE, spectro_dir = file.path(tempdir(), "Spectros"), time_scale = 1, 
  ticks = c(1000, 10000, 1000) # Tick marks from 1 to 10 kHz with 1 kHz interval
) 

# Take a look at the spectrograms and compare them with the previous extraction.
nrow(TD$data$event_data) # Only three audio events!

CATBIC <- read_audio(file.path(data_dir, "Catharus-bicknelli-54864.mp3"))
TD <- threshold_detection(
  CATBIC, threshold = 12, time_exp = 1, min_dur = 140, max_dur = 440, min_TBE = 10, 
  max_TBE = Inf, EDG = 0.996, LPF = 10000, HPF = 1000, FFT_size = 256, FFT_overlap = 0.875, 
  start_thr = 22, end_thr = 30, SNR_thr = 10, angle_thr = 125, duration_thr = 440, NWS = 1000,
  KPE = 1e-05, KME = 1e-05, settings = FALSE, acoustic_feat = TRUE, metadata = FALSE
)

# Acoustic features are stored in a data frame called event_data,
# stored by order of detection.

View(TD$data$event_data) # Contains the filename and the time of detection in the 
                         # recording, and 26 extracted features.

# Start and end of the 5th extracted audio event (in samples)
c(TD$data$event_start[[5]], TD$data$event_end[[5]])

# Remember you just have to divide by the sample rate to retrieve the time (s)
c(TD$data$event_start[[5]], TD$data$event_end[[5]]) / slot(CATBIC, "samp.rate")

par(mar = c(1,1, 1, 1), oma = c(1, 1, 1, 1))

# Amplitude track of the 5th audio event
plot(TD$data$amp_track[[5]], type = "l")

# Frequency track of the 5th audio event
plot(TD$data$freq_track[[5]], type = "l")

# Access the arguments of the blob_detection function
?blob_detection

# Use the bat recording stored in the package
data(myotis)

# Set each argument according to the targeted audio events
BD <- blob_detection(
  myotis, # Either a path to an audio file (see ?read_audio), or a Wave object
  time_exp = 10, # Time expansion factor of 10 for time expanded recordings.
  min_dur = 1.5, # Minimum duration threshold of 1.5 milliseconds (ms)
  max_dur = 80, # Maximum duration threshold of 80 ms
  min_area = 40, # minimum number of 40 pixels in the blob
  min_TBE = 20, # Minimum time window between two audio events of 20 milliseconds
  EDG = 0.996, # Temporal masking with Exponential Decay Gain from 0 to 1
  LPF = slot(myotis, "samp.rate") * 10 / 2, # Low-Pass Filter at the Nyquist frequency
  HPF = 16000, # High-Pass Filter of 16 kHz
  FFT_size = 256, # Size of the Fast Fourrier Transform (FFT) window
  FFT_overlap = 0.875, # Percentage of overlap between two FFT windows

  blur = 2, # Gaussian smoothing function with a factor of 2 for blurring the spectrogram
  bg_substract = 20, # Foreground extraction with a mean filter applied on the spectrogram
  contrast_boost = 20, # Edge contrast enhancement filter of the spectrogram contour

  settings = FALSE, #  Save on a list the above parameters set with this function
  acoustic_feat = TRUE, # Extracts the acoustic and signal quality parameters 
  metadata = FALSE, # Extracts on a list the metadata embedded with the Wave file
  spectro_dir = file.path(tempdir(), "Spectros"), # HTML page with spectrograms by order of detection 
  time_scale = 0.1, # Time resolution of 2 ms for spectrogram display
  ticks = TRUE # Tick marks and their intervals are drawn on y-axis (frequencies)
) 

# Get the number of extracted audio events
nrow(BD$data$event_data)

# Let's try various settings, starting with 512 FFT size instead of 256.
BD <- blob_detection(
  myotis, time_exp = 10, FFT_size = 512, settings = FALSE, acoustic_feat = TRUE,
  metadata = FALSE, spectro_dir = file.path(tempdir(), "Spectros"), time_scale = 0.1, ticks = TRUE
) 

# Take a look at the spectrograms and compare them with the previous extraction.
nrow(BD$data$event_data) # Only 6 audio events!

# Acoustic features
head(BD$data)