1. Recognition Process (HTK)
Tools
Lexicon
Syntactic Network
“phonemes”
Input:Speech
recognition
Output: Text

2. Phonemes A set of (typically) 40-50 phonemic symbols
symbols (TIMIT/SAMPA) conforming to IPA
Language-dependent
Modelled in stochastic models (HMMs)
Trained on large annotated speech databases
Additional silence and garbage models (for rejection)

3. Lexicon (HTK) phonemic tr. Orthographic forms
A ah sp
A ax sp
A ey sp
CALL k ao l sp
DIAL d ay ax l sp
EIGHT ey t sp
PHONE f ow n sp
SENT-END [] sil
SENT-START [] sil
SEVEN s eh v n sp
TO t ax sp
TO t uw sp
ZERO z ia r ow sp
phonemic tr.
Orthographic
forms
output string if not the orth. form

4. Syntactic Network (HTK)

5. Expansion of structures 1

6. Expansion of structures 2

7. HTK Standard Lattice HTK tools GrapHvite NetBuilder CPK NLP Suite
1) Implement grammar in HTK (EBNF-like) grammar definition language
2) Use HParse on grammar to create HTK Standard Lattice
3) Create word list (no tool for deriving WL from grammar!)
4) Use HDMan to create Dictionary with transcriptions (presupposes pronunciation dictionary)
GrapHvite NetBuilder
1) “Draw” the network(s)
2) export word list and add transcriptions
CPK NLP Suite

8. Making a grammar network (HTK)
HTK-Tools

9. Using the HTK Grammar Tools
An EBNF-like grammar definition language
HTK tools: Hparse, Hbuild, HSGen, HDMan
$digit = ONE | TWO | THREE | FOUR | FIVE |
SIX | SEVEN | EIGHT | NINE | OH | ZERO;
$name = [ JOOP ] JANSEN |
[ JULIAN ] ODELL |
[ DAVE ] OLLASON |
[ PHIL ] WOODLAND |
[ STEVE ] YOUNG;
( SENT-START ( DIAL <$digit> | (PHONE|CALL) $name) SENT-END )

10. Using GrapHvite Grammar Tool
Visual Netbuilder:

11. GrapHvite NetBuilder
# Define size of network: N=num nodes and L=num arcs
N=4 L=8
# List nodes: I=node-number, W=word
I=0 W=start
I=1 W=end
I=2 W=bit
I=3 W=but
# List arcs: J=arc-number, S=start-node, E=end-node
J=0 S=0 E=2
J=1 S=0 E=3
J=2 S=3 E=1
J=3 S=2 E=1
J=4 S=2 E=3
J=5 S=3 E=3
J=6 S=3 E=2
J=7 S=2 E=2

12. CPK Suite

13. Testing/running mygram.aps
testing on a test database
apspars mygram.aps mygram.snt
testing random generation
apssgen mygram.aps
simulating recognition
apstrec mygram.aps 0
simulating speech understanding
apstslu mygram.aps 0
fullgram HTK standard lattice
apsconv -f -h mygram.aps