I have just realised I missed a few comments out of my earlier message.
1041 <likeness>: YT spots the suffix if it is parsed as -n@s. I've
changed this to -nIs in the .pho file so that it more closely resembles
the labelled file in the database.
1093 <roasting>: YT parses the suffix as -INg (for overlap purposes; I
have deleted the value for [g], leaving -IN).
1096 <postman>: label file has no [t], though the structure implies it, so
I've inserted a t with duration 0 in the file 1096MH.pho.
1117 <resting>: as 1093
1176 <buckle>: NB YT has [@] with duration 0 (because of overlap).
1180 <witness>: as 1041.
Paul