divime_sad: run a DiViMe SAD module
In gobbios/avutils: various utilities for DiViMe tools and audio processing

Description Usage Arguments Details Value

run a DiViMe SAD module

1
2
3

divime_sad(audio_loc, divime_loc, module = "noisemes",
  splitaudio = NULL, vmstart = TRUE, vmshutdown = TRUE,
  messages = TRUE, overwrite = FALSE)

`audio_loc`	character, path to the audio files
`divime_loc`	character, path to the DiViMe directory with a VM
`module`	character, which module to execute (default is `"noisemes"`), see details
`splitaudio`	numeric, should audio files be split into smaller chunks before processing by SAD tool, default is `NULL`, see details
`vmstart`	logical, perform a check whether the VM is running and if not start it up (by default `TRUE`). Turning this off, will speed up the function a little bit, but requires that you are sure that the VM is indeed running in `divime_loc`.
`vmshutdown`	logical, should the VM shut down after the operations are done (by default `TRUE`)
`messages`	logical, should the file names of each processed file be printed
`overwrite`	logical, should output files be overwritten if they already exist (default is `FALSE`)

module= sets the SAD module to be used: can be either "noisemes", "opensmile" or "tocombo"

It appears that some of the modules have difficulties with larger audio files (opensmile and noisemes). Hence, setting splitaudio= to a numeric value will temporarilly split the source audio into chunks of that duration (split_audio). Im my experience, a chunk duration of about two minutes solves these issues (e.g. via splitaudio=120). Note that this step requires the sox utility available (see set_binaries and split_audio).

a data.frame with the locations of the created rttm files and some diagnostics

gobbios/avutils documentation built on Feb. 19, 2020, 9:44 a.m.