Automatic Speech Recognition
Introduction
Readings: Jurafsky & Martin 7.1

2
HLT Survey Chapter 1
The Human Dialogue System
The Human Dialogue System
Computer Dialogue Systems
Audition
Automatic
Speech
Recognition
Natural
Language
Understanding
Dialogue
Management
Planning
Natural
Language
Generation
Text

to

speech
signal
words
logical form
words
signal
signal
Computer Dialogue Systems
Audition
ASR
NLU
Dialogue
Mgmt.
Planning
NLG
Text

to

speech
signal
words
logical form
words
signal
signal
Parameters of ASR Capabilities
•
Different types of tasks with different difficulties
–
Speaking mode (isolated words/continuous speech)
–
Speaking style (read/spontaneous)
–
Enrollment (speaker

independent/dependent)
–
Vocabulary (small < 20 wd/large >20kword)
–
Language model (finite state/context sensitive)
–
Perplexity (small < 10/large >100)
–
Signal

to

noise ratio (high > 30 dB/low < 10dB)
–
Transducer (high quality microphone/telephone)
The Noisy Channel Model
message
noisy channel
message
Message
Channel
+
=Signal
Decoding model: find Message*= argmax P(MessageSignal)
But how do we represent each of these things?
ASR using HMMs
•
Try to solve P(MessageSignal) by breaking the
problem up into separate components
•
Most common method:
Hidden Markov Models
–
Assume that a message is composed of words
–
Assume that words are composed of sub

word parts
(phones)
–
Assume that phones have some sort of acoustic
realization
–
Use probabilistic models for matching acoustics to
phones to words
HMMs: The Traditional View
go
home
g
o
h
o
m
x
0
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
Markov model
backbone composed
of phones
(hidden because we
don’t know
correspondences)
Acoustic observations
Each line represents a probability estimate (more later)
HMMs: The Traditional View
go
home
g
o
h
o
m
x
0
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
Markov model
backbone composed
of phones
(hidden because we
don’t know
correspondences)
Acoustic observations
Even with same word hypothesis, can have different alignments.
Also, have to search over all word hypotheses
HMMs as Dynamic Bayesian
Networks
go
home
q
0
=g
x
0
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
Markov model
backbone composed
of phones
Acoustic observations
q
1
=o
q
2
=o
q
3
=o
q
4
=h
q
5
=o
q
6
=o
q
7
=o
q
8
=m
q
9
=m
HMMs as Dynamic Bayesian
Networks
go
home
q
0
=g
x
0
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
Markov model
backbone composed
of phones
q
1
=o
q
2
=o
q
3
=o
q
4
=h
q
5
=o
q
6
=o
q
7
=o
q
8
=m
q
9
=m
ASR: What is best assignment to q
0
…q
9
given x
0
…x
9
?
Hidden Markov Models & DBNs
DBN representation
Markov Model
representation
Parts of an ASR System
Feature
Calculation
Language
Modeling
Acoustic
Modeling
k
@
Pronunciation
Modeling
cat: k@t
dog: dog
mail: mAl
the: D&, DE
…
cat dog: 0.00002
cat the: 0.0000005
the cat: 0.029
the dog: 0.031
the mail: 0.054
…
The cat chased the dog
S E A R C H
Parts of an ASR System
Feature
Calculation
Language
Modeling
Acoustic
Modeling
k
@
Pronunciation
Modeling
cat: k@t
dog: dog
mail: mAl
the: D&, DE
…
cat dog: 0.00002
cat the: 0.0000005
the cat: 0.029
the dog: 0.031
the mail: 0.054
…
Produces
acoustics (x
t
)
Maps acoustics
to phones
Maps phones
to words
Strings words
together
Feature calculation
Feature calculation
Frequency
Time
Find energy at each time step in
each frequency channel
Feature calculation
Frequency
Time
Take inverse Discrete Fourier
Transform to decorrelate frequencies
Feature calculation

0.1
0.3
1.4

1.2
2.3
2.6
…
0.2
0.1
1.2

1.2
4.4
2.2
…

6.1

2.1
3.1
2.4
1.0
2.2
…
0.2
0.0
1.2

1.2
4.4
2.2
…
…
Input:
Output:
Robust Speech Recognition
•
Different schemes have been developed for
dealing with noise, reverberation
–
Additive noise: reduce effects of particular
frequencies
–
Convolutional noise: remove effects of linear
filters (cepstral mean subtraction)
Now what?

0.1
0.3
1.4

1.2
2.3
2.6
…
0.2
0.1
1.2

1.2
4.4
2.2
…

6.1

2.1
3.1
2.4
1.0
2.2
…
0.2
0.0
1.2

1.2
4.4
2.2
…
That you …
???
Machine Learning!

0.1
0.3
1.4

1.2
2.3
2.6
…
0.2
0.1
1.2

1.2
4.4
2.2
…

6.1

2.1
3.1
2.4
1.0
2.2
…
0.2
0.0
1.2

1.2
4.4
2.2
…
That you …
Pattern recognition
with HMMs
Hidden Markov Models (again!)
P(acoustics
t
state
t
)
Acoustic Model
P(state
t+1
state
t
)
Pronunciation/Language models
Acoustic Model

0.1
0.3
1.4

1.2
2.3
2.6
…
0.2
0.1
1.2

1.2
4.4
2.2
…

6.1

2.1
3.1
2.4
1.0
2.2
…
0.2
0.0
1.2

1.2
4.4
2.2
…
dh
a
a
t
•
Assume that you can
label each vector with
a phonetic label
•
Collect all of the
examples of a phone
together and build a
Gaussian model (or
some other statistical
model, e.g. neural
networks)
N
a
(
m,S
)
P(Xstate=a)
Building up the Markov Model
•
Start with a model for each phone
•
Typically, we use 3 states per phone to give
a minimum duration constraint, but ignore
that here…
a
p
1

p
transition probability
a
p
1

p
a
p
1

p
a
p
1

p
Building up the Markov Model
•
Pronunciation model gives connections
between phones and words
•
Multiple pronunciations:
ow
t
m
dh
p
dh
1

p
dh
a
p
a
1

p
a
t
p
t
1

p
t
ah
ow
ey
ah
t
Building up the Markov Model
•
Language model gives connections between
words (e.g., bigram grammar)
dh
a
t
h
iy
y
uw
p(hethat)
p(youthat)
ASR as Bayesian Inference
q
1
w
1
q
2
w
1
q
3
w
1
x
1
x
2
x
3
th
a
t
h
iy
y
uw
p(hethat)
p(youthat)
h
iy
sh
uh
d
argmax
W
P(WX)
=argmax
W
P(XW)P(W)/P(X)
=argmax
W
P(XW)P(W)
=argmax
W
S
Q
P(X,QW)P(W)
≈argmax
W
max
Q
P(X,QW)P(W)
≈argmax
W
max
Q
P(XQ) P(QW) P(W)
ASR Probability Models
•
Three probability models
–
P(XQ): acoustic model
–
P(QW): duration/transition/pronunciation
model
–
P(W): language model
•
language/pronunciation models inferred
from prior knowledge
•
Other models learned from data (how?)
Parts of an ASR System
Feature
Calculation
Language
Modeling
Acoustic
Modeling
k
@
Pronunciation
Modeling
cat: k@t
dog: dog
mail: mAl
the: D&, DE
…
cat dog: 0.00002
cat the: 0.0000005
the cat: 0.029
the dog: 0.031
the mail: 0.054
…
The cat chased the dog
S E A R C H
P(XQ)
P(QW)
P(W)
EM for ASR: The Forward

Backward Algorithm
•
Determine “state occupancy” probabilities
–
I.e. assign each data vector to a state
•
Calculate new transition probabilities, new
means & standard deviations (emission
probabilities) using assignments
ASR as Bayesian Inference
q
1
w
1
q
2
w
1
q
3
w
1
x
1
x
2
x
3
th
a
t
h
iy
y
uw
p(hethat)
p(youthat)
h
iy
sh
uh
d
argmax
W
P(WX)
=argmax
W
P(XW)P(W)/P(X)
=argmax
W
P(XW)P(W)
=argmax
W
S
Q
P(X,QW)P(W)
≈argmax
W
max
Q
P(X,QW)P(W)
≈argmax
W
max
Q
P(XQ) P(QW) P(W)
Search
•
When trying to find W*=argmax
W
P(WX), need
to look at (in theory)
–
All possible word sequences W
–
All possible segmentations/alignments of W&X
•
Generally, this is done by searching the space of
W
–
Viterbi search: dynamic programming approach that
looks for the most likely path
–
A* search: alternative method that keeps a stack of
hypotheses around
•
If W is large, pruning becomes important
How to train an ASR system
•
Have a speech corpus at hand
–
Should have word (and preferrably phone)
transcriptions
–
Divide into training, development, and test sets
•
Develop models of prior knowledge
–
Pronunciation dictionary
–
Grammar
•
Train acoustic models
–
Possibly realigning corpus phonetically
How to train an ASR system
•
Test on your development data (baseline)
•
**Think real hard
•
Figure out some neat new modification
•
Retrain system component
•
Test on your development data
•
Lather, rinse, repeat **
•
Then, at the end of the project, test on the test
data.
Judging the quality of a system
•
Usually, ASR performance is judged by the
word error rate
ErrorRate = 100*(Subs + Ins + Dels) / Nwords
REF: I WANT TO GO HOME ***
REC: * WANT TWO GO HOME NOW
SC: D C S C C I
100*(1S+1I+1D)/5 = 60%
Judging the quality of a system
•
Usually, ASR performance is judged by the
word error rate
•
This assumes that all errors are equal
–
Also, a bit of a mismatch between optimization
criterion and error measurement
•
Other (task specific) measures sometimes
used
–
Task completion
–
Concept error rate
Comments 0
Log in to post a comment