1. 語音訊號處理之初步實驗 NTU Speech Lab 指導教授: 李琳山 助教: 熊信寬
ANN Tutorial
語音訊號處理之初步實驗
NTU Speech Lab
指導教授: 李琳山
助教: 熊信寬

2. ㄊㄞ ㄨㄢ ㄉㄚ ㄒㄩ ㄝ
Acoustic Model
Speech Recognition
Language Model 台 灣 大 學

3. How do we build this machine?
ㄅ ㄆ ㄇ ㄈ
ㄉ ㄊ ㄋ ㄌ
ㄍ ㄎ ㄏ
ㄐ ㄑ ㄒ
ㄓ ㄔ ㄕ ㄖ
ㄗ ㄘ ㄙ
ㄧ ㄨ ㄩ
ㄚ ㄛ ㄜ ㄝ
ㄞ ㄟ ㄠ ㄡ
ㄢ ㄣ ㄤ ㄥ ㄦ
Classifier
How do we build this machine?
Our Task

4. “Field of study that gives computers the ability to learn without being explicitly programmed”
Instead of writing a program by hand for each specific task, we collect lots of examples that specify the correct output for a given input.
A machine learning algorithm then takes these examples and produces a program that does the job.
The program produced by the learning algorithm may look very different from a typical hand-written program. It may contain millions of numbers.
If we do it right, the program works for new cases as well as the ones we trained it on.
If the data changes the program can change too by training on the new data.
Massive amounts of computation are now cheaper than paying someone to write a task-specific program.
Instead of writing a program by hand for each specific task, we collect lots of examples that specify the correct output for a given input.
A machine learning algorithm then takes these examples and produces a program that does the job.
The program produced by the learning algorithm may look very different from a typical hand-written program. It may contain millions of numbers.
If we do it right, the program works for new cases as well as the ones we trained it on.
If the data changes the program can change too by training on the new data.
Massive amounts of computation are now cheaper than paying someone to write a task-specific program.
Machine Learning
Reference: Neural Networks for Machine Learning course by Geoffrey Hinton, Coursera

5. The Classification Task
Features
Classifier
Classes
Male
Hair Length
Make-up .
Classifier
Female
Others
舉例而言 若要針對這房間裡的人 做性別的分類
The Classification Task

6. Male
Female
Make-Up
2D Feature Space
Hair Length
Voice pitch

7. We need some type of non-linear function!
Multi-D Feature Space

8. Neurons Each neuron receives inputs from other neurons
The effect of each input line on the neuron is controlled by a synaptic weight
The weights can be positive or negative.
The synaptic weights adapt so that the whole network learns to perform useful computations
Recognizing objects, understanding language, making plans, controlling the body.
Neurons
Reference: Neural Networks for Machine Learning course by Geoffrey Hinton, Coursera

9. Artificial Neural Network
Feed Forward Net
activation function
bias
i input th x1 w1 x2 w2
output
index over
input connections
weight on x3 w3 y th i
input w4 x4 b 1
Artificial Neural Network
A lot of simple non-linearity  complex non-linearity

10. Sigmoid Function A.K.A Logistic Function
Activation Function

11. How Do Neural Nets Learn?
Intuition:
1. Start with random weights
2. Compare the outputs of the net to the targets
3. Try to adjust the weights to match the outputs with the targets yj tj
Target 1
0.2 4
0.9 1 -3 w
How Do Neural Nets Learn?

12. Gradient Descent x1 w1 x2 w2 x3 w3 x4 w4 Error w2 w1 Updated weights
Learning rate
Reference: Machine Learning course by Andrew Ng, Coursera

13. Back Propagation yj j zj i yi
Reference: Neural Networks for Machine Learning course by Geoffrey Hinton, Coursera

14. Overfitting Which model do you trust?
The complicated model fits the data better.
But it is not economical.
A model is convincing when it fits a lot of data surprisingly well.
Not just the training data.
output = y
input = x
Which output value should you predict for this test input?
Overfitting
Reference: Neural Networks for Machine Learning course by Geoffrey Hinton, Coursera

15. Divide dataset into training set and validation set
Use training set for training, validation set as indicator to overfitting
Use another independent testing test to performance
Training and Testing

16. ㄊㄞ ㄨㄢ ㄉㄚ ㄒㄩ ㄝ
Acoustic Model
Speech Recognition
Language Model 台 灣 大 學

17. Our Task ㄅ ㄆ ㄇ ㄈ ㄉ ㄊ ㄋ ㄌ ㄍ ㄎ ㄏ ㄐ ㄑ ㄒ ㄓ ㄔ ㄕ ㄖ ㄗ ㄘ ㄙ ㄧ ㄨ ㄩ ㄚ ㄛ ㄜ ㄝ
ㄞ ㄟ ㄠ ㄡ
ㄢ ㄣ ㄤ ㄥ ㄦ
Our Task

18. We can ‘splice’ the right and left 4 MFCCs to form a 13 x 9 = 117 Dim Vector
Example --- Inputs
MFCC:13 Dim Vector

19. Example --- Output Right context dependent initial final phones
‘a’, ‘ai’, ‘an’, ‘ang’, ‘au’, ‘b_a’, ‘b_e’, ‘b_ee’, ‘b_i’ …
Target
‘a’
‘ai’ 1
117
Dim
MFCC
‘an’
Example --- Output

20. Matlab NN toolkit---Net
net = newff(Feature,Label,10);
% given Feature and Label, this command generates an NN with 10 hidden units
net = newff(Feature,Label,[20 10],{'logsig' 'tansig'});
% this command generates an NN with two layers, one with 20 units and sigmoid as activation function, one with 10 units and tanh as activation function
Matlab NN toolkit---Net

21. Matlab NN Toolkit ---Train
[net,tr] = train(net,Feature,Label);
% this command will train the network, dividing your dataset into training, validation, and testing set
Matlab NN Toolkit ---Train

22. Matlab NN Toolkit ---Test
out = sim(net,testInputs);
% this command runs the command you have trained on a test data set to obtain outputs
Matlab NN Toolkit ---Test

23. Matlab NN Toolkit ---NNTool
% this command calls a GUI for NN visualization
Matlab NN Toolkit ---NNTool

24. Demo Program Location: speech.ee.ntu.edu.tw/~poetek/nnkit
Copy the folder to your computer, and run the file ‘run.m’ for a quick demo
Demo Program

25. DEMO

26. Demo Program --- Results
Example (1): Classify the inputs into 4 classes: {‘a’, ‘e’, ‘i’, ‘u’}
ans =
correct: 1139 out of 1205
accuracy =
0.9452
classification =
Demo Program --- Results

27. Demo Program --- Results(2)
Example (2): Classify the inputs into 6 classes: {‘a’, ‘e’, ‘i’, ‘u’, ‘b_a’, ‘p_a’, ‘m_a’, ‘f_a’}
correct: 1293 out of 1446
accuracy =
0.8942
classification =
Demo Program --- Results(2)

28. Demo Program --- Results(3)
Example (3): Classify the inputs into all 147 classes
correct: 92 out of 506
accuracy =
0.1818
Demo Program --- Results(3)

29. Please run example 1~3 with the demo program, record your performance, run time and your observations
Please play around with different settings to achieve better accuracy
Some parameters you can change:
Net structure
Training algorithm
Input Features
net.trainParam
(Please type ‘help newff’ for more info)
Assignments

30. Neural Networks for Machine Learning course by Geoffrey Hinton, Coursera
Machine Learning course by Andrew Ng, Coursera
Matlab NN toolkit documentation
References

31. Thank You! Feel free to ask questions!My
Thank You!