More transcription script improvements

bin/transcribe-audio

@@ -1,5 +1,12 @@
 #!/usr/bin/env bash
 
+# The base model is pretty good overall. It has good punctuation inserting and catches most words.
+# Tiny is fast and often has correct grammar, but it misses a lot of words, especially when the
+# source isn't loud or is muffled.
+#
+# Small and medium models can do better word detection at times, but suffer from a lack of punctuation.
+# Medium is particularly bad and often excludes periods and commas.
+
 if which tput >/dev/null 2>&1; then
     ncolors=$(tput colors)
 fi