New automated phoneme extractor for Faceposer -- works on Vista/7

courtesy of the hlcoders mailing list

http://steamreview.org/external/vdc/phonemeextractor_sapi5.4.zip

This updated extractor provides better results than the SDK’s default one. It works on Windows 7, and should work on Vista too. It will not work on XP.

To install, extract the contents of the zip file to [SDK version]\bin\phonemeextractors. KEEP A BACKUP HANDY, as every time you run the SDK Launcher, or in the case of L4D1/2 receive an update, the custom DLL will be replaced with Valve’s original and you’ll have to put it
back in place manually.

The easiest way to run Faceposer without using Valve’s launcher is with
the Source SDK Windows Gadget: http://developer.valvesoftware.com/wiki/Source_SDK_Windows_gadget

This will make Faceposer crash if your sentence text contains special characters (e.g. commas and apostrophes) when trying to extract

This is fantastic

Holy shit, awesome.

note that it doesn’t give phonemes for me, but it places words automatically and quite accurately, which is already a good gain of time

this is how it behaves here

http://dl.dropbox.com/u/3797350/img/automatedphonemevista.png

the words are placed very accurately (even though the detection completely fails on longer sentences, so I’d recommend splitting up your WAVs), and all you’ve gotta do is select the first phoneme, right click, and “delete all phonemes after this one”

and possibly delete the duplicate words it sometimes create (like the extra “ago” at the end)

wow brilliant find max.

Another small warning: on one of my computers, this causes Faceposer to crash on long lines and whenever you try to resize a word or a phoneme

If you have a lot of lines to sync, i’d just recommend to extract all the words from all the lines then going through all of them with the regular DLL again

About time.

It does not work for me. Halp ):

Its not working for me, I’m on Windows 7 Ultimate 64bit.

maybe its for 32-bit?

no, it works on both my machines

I finally got around to doing it, it says “missing phonemes, logging to desktop” and it does the same thing it does with Max of S2D’s faceposer.

I love you

[editline]12:16AM[/editline]

Fuck doesn’t work on Vista x64.

Put it in SDK/bin/phonemeextractors/ and SDK/Source2007/bin/phonemeextractors

:frowning:

[editline]12:28AM[/editline]

All my hopes and dreams are now dead

Finally… I’ve been waiting for this.

I got to the point of word recognition but not phoneme like Max. Kinda pointless :expressionless:

Well not entirely pointless but I think I’ll wait until it can fill in the phonemes

uh, it’s not pointless at all

you do realize that once the word timing is all set (which would take a LOT of time without the extractor), just right click on a word, insert phonemes, and enter your phonemes

90% of the time if you do it right you won’t have to readjust the phoneme timing, which makes it so that you can lipsync a ten second voice file in approximately a minute

Or you can get to the point of inserting phonemes, try to resize a phoneme, and crash. Every. Time.

my point is, I have 50 wavs to do

what I can do is lipsync everything then come back without the DLL to do adjustments on the 2 or 3 ones that had problems

as a whole it stays faster

And if you’re doing things in a minute it will probably end up being crap

You know, I never really liked automatic extractors before, but that one does the only task that is really repetitive, boring and tedious: placing words

I see absolutely no reason to complain here