[Sarah B]

I’m confused about the use of HERest in the first re-estimation step of the do_alignment script. The script says that this is re-estimating model parameters, is this giving them some initial values beyond what they were when the HMMs were constructed? I would have expected to start with Viterbi (HVite)

[Sarah B]

Ignore this – posted in wrong place, sorry!

[Sarah B]

I got the “SIOD ERROR: wrong type of argument to get_c_val” with no backtrace when the script got to lines of my data that started with ! or ). Removing those characters from the start of lines allowed me to run the script.

[Sarah B]

From the answer given, I’m still unclear what answer the question is looking for. The options given are:
a. i. and iii. only
b. ii. only
c. ii., iii. and iv.
d. iv. only

i. is false
Regarding ii, I’m not sure what this would mean. Is it a trick question that we might misread as talking about a spectrogram with no sudden/unnatural changes?
If we are talking about using only linear prediction, I would go for d., but if we are talking about combining linear prediction with a source sound, then I would go for c., because then iii and iv seem to be true.

[Sarah B]

On p.355 Taylor mentions r[n] referring to radiation. In section 11, he describes this as “radiation impedance from the lips”. Why is this separated out, rather than considered part of the vocal-tract filter?

[Author]

Posts