hi,
Flo today forwarded me the mail of an opensmile user who was wondering about high pitch outliers. While reconstructing the issue I observed higher pitch outlier values for opensmile eGeMAPSv01b LLDs than for ComParE LLDs. This is most likely due to a missing maxPitch constraint in the eGeMAPSv01b_core.lld.conf.inc. One can live with this of course. Only wanted to make sure that this missing maxPitch constraint is intended.
Furthermore, for eGeMAPSv01b as opposed to ComParE no voicingCutoff is specified. Do you know, whether the default used by the binary is the same as specified in the ComParE config ? This information I'd like to forward to the opensmile user.
Thanks!
hi,
Flo today forwarded me the mail of an opensmile user who was wondering about high pitch outliers. While reconstructing the issue I observed higher pitch outlier values for opensmile
eGeMAPSv01bLLDs than forComParELLDs. This is most likely due to a missingmaxPitchconstraint in the eGeMAPSv01b_core.lld.conf.inc. One can live with this of course. Only wanted to make sure that this missingmaxPitchconstraint is intended.Furthermore, for
eGeMAPSv01bas opposed toComParEnovoicingCutoffis specified. Do you know, whether the default used by the binary is the same as specified in the ComParE config ? This information I'd like to forward to the opensmile user.Thanks!