-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
523 lines (307 loc) · 13 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
<!DOCTYPE html>
<html>
<head>
<title>Title</title>
<meta charset="utf-8">
<style>
@import url(https://fonts.googleapis.com/css?family=Yanone+Kaffeesatz);
@import url(https://fonts.googleapis.com/css?family=Droid+Serif:400,700,400italic);
@import url(https://fonts.googleapis.com/css?family=Ubuntu+Mono:400,700,400italic);
body { font-family: 'Droid Serif'; }
h1, h2, h3 {
font-family: 'Yanone Kaffeesatz';
font-weight: normal;
}
.remark-code, .remark-inline-code { font-family: 'Ubuntu Mono'; }
</style>
</head>
<body>
<textarea id="source">
# Voice Recognition - Breif Theory
Turning speech into text using *Automatic Speech Recognition* (**ASR**)
- Uses machine learning to turn voice input into textual output
- Machine learning models based on voice input and textual meaning
---
# Voice Recognition - Breif Theory
Turn text into something actionable using *Natural Language Processing* (**NLP**) specifically *Natural Language Understanding* (**NLU**)
- Syntactic and semantic processing
- Text converted to intents and entities (synonyms)
**Example**: "I would like to fly from DCA to Chicago"
**Intent**: FindFlightIntent
**Entities**: DCA, Chicago
---
# Voice Recognition - Breif Theory
Natural Language Understanding is more than just parsing. One example is the application of synonyms.
**Example**: "I would like to fly from DCA to MDW"
- MDW is the same as Chicago
**Example**: "I would like to fly from Washington, DC to MDW"
- Washington, DC is the same as DCA
**Example**: "I would like to fly from Reagan to Chicago"
- DCA is the same as Reagan International and even just Reagan
Disambiguation can be difficult
- Washington, DC could be DCA or IAD
- Chicago could be Midway, MDW, O'Hare or ORD
---
# Voice Recognition - Breif Theory
Responses can also benifit from NLP with using *Natural Language Generation* (**NLG**)
- Pull in context around a given response instead of simplely generating a response based off a state machine
- If you are interested in learning more check out Automated Insights' [Wordsmith](https://automatedinsights.com/wordsmith)
---
# Voice Recognition - Landscape
**Main Players**: Siri, Google, Cortana, Alexa
- Each has a slightly different integration
- Some are fenced like Siri with a limited set of interactions allowed
- Alexa and Google are leaders
---
# Voice Recognition - Landscape
Evolution of Alexa from Echo to Echo Dot to Echo Show and more
---
# Voice Recognition - Landscape
Amazon has also created the ability for hardware manufacturers to create their own device so Alexa can be integrated into non-Amazon devices
---
# Voice Recognition - Landscape
There are two acronyms you will see when looking at the Alexa developer documentation.
*Alexa Voice Service* (**AVS**)
- This service enables the hardware integrations
- The documentation for [Alexa Voice Service](https://developer.amazon.com/docs/alexa-voice-service/)
*Alexa Skills Kit* (**ASK**)
- This service enables the development of skills for Alexa
- The documentation for [Alexa Skills Kit](https://developer.amazon.com/docs/ask-overviews/build-skills-with-the-alexa-skills-kit.html)
---
# Voice Recognition - Landscape
Amazon has also extracted two services that back Alexa that can be used independantly.
*AWS Lex*
- Exposes the NLU system that backs Alexa
- Documentation for [AWS Lex](https://aws.amazon.com/lex/)
*AWS Polly*
- Exposes the speech synthesis that backs Alexa
- Documentation for [AWS Polly](https://aws.amazon.com/polly/)
---
# Alexa Details - Alexa processing flow
Specifics of how Alexa uses voice recognition to process a request.
1. User speaks
2. Alexa device detects wake word
3. Alexa device sends audio stream to the Alexa service
4. Alexa service applies ASR and NLU
5. The invocation word is used to find the correct skill to call
6. Your skill is invoked with the extracted Intent and Slots with their Entities
7. A response is returned from the skill back to the Alexa service and streamed as speach back to the Alexa device
---
# Alexa Details - Skill development process
1. Design a voice UI for the skill
2. Gather the intents, utterances and slots needed for the skill
3. Implementation
4. Deploy and test
---
# Alexa Details - Advanced concepts
- Contextual understanding - Examples would be using the previous request to understand the next request
or using the user's location to disambiguate a location.
- Multimodal contextual understanding (video + voice, etc)
- Alexa knows the difference between multiple speakers, a feature called "Your Voice"
- "Natural skill interaction" allows dropping of the invocation name if the request
context can be determined to match only one skill
---
# Skill Development - Basics
AWS Lambda is the prefered method of skill processing. Most of the examples you will find
are written in Javascript but it isn't the only option available.
- Another option is to use Python and [Flask Ask](https://github.com/johnwheeler/flask-ask)
The Alexa development portal is a seperate area from the AWS Lambda portal. There is a non-trival
process that is required to let the Alexa system communicate with a Lambda application, more about
that can be found in the [Custom Skills documentation](https://developer.amazon.com/docs/custom-skills/host-a-custom-skill-as-an-aws-lambda-function.html)
- Note that we will be using Heroku because it is free and easier to get started with.
---
# Skill Development - Intents
There are both custom and built in intents available to Skills.
- A custom intent is given a name and that name will later be referenced in the code for the
Skill. An example might be:
```GetHoroscope```
- There are a number of built in intents that may be used to reduce the amount of work needed
in setting up common requests. A couple examples are:
- AMAZON.HelpIntent - A request for help
- AMAZON.StopIntent - A request to stop the skill
More can be found in the [built in intents document](https://developer.amazon.com/docs/custom-skills/standard-built-in-intents.html)
---
# Skill Development - Slots
There are both custom and built in slots available to Skills.
- A custom slot will consist of a list of valid slot values. These values will be used to produce a model
for the slot. For example you might create a custom slot type named *sign* as follows:
```
Capricorn
Aquarius
Pisces
Aries
Taurus
Gemini
```
- There are a number of built in slot types that will fit a large number of needs. A couple examples are:
- AMAZON.DATE - A slot that contains a date
- AMAZON.NUMBER - A slot that contains a number
More can be found in the [built in slot types document](https://developer.amazon.com/docs/custom-skills/slot-type-reference.html)
---
# Skill Development - Sample Utterances
- Utterances help form the model for recognizing intents based on your intents
and slots
- The more sample utterances you have the more likely the model will be able to
match a phrase that isn't exactly what is in one of the utterances
- An example of a set of utterances:
```
GetHoroscope the horoscope for {Sign}
GetHoroscope {Sign}
GetHoroscope what's the horoscope for {Sign}
GetHoroscope what is the horoscope for {Sign}
GetHoroscope give me the horoscope for {Sign}
GetHoroscope tell me the horoscope for {Sign}
```
---
# Skill Development - Responses
The response text that you give back to the Alexa skills service as a response to
a request can generally be translated into speech well. In some cases you may want
to enhance it with one of the following options.
- SSML - Speech Synthesis Markup Language allows you to provide extra information
to the text to speech system such as reading a list of numbers as a single number
or slowing down a set of words. An example would be:
```
<speak>
<say-as interpret-as="cardinal">12345</say-as>
</speak>
```
- SSML also lets you send back pre-recorded audio. An example would be:
```
<speak>
<audio src="https://something.com/audio/welcome.mp3" />
</speak>
```
---
class: middle, center
# Questions?
---
# Voice UX/UI - General
There are a number of places to apply UX/UI processes in designing an Alexa Skill.
The first two to know about are:
- Voice-Only Interaction - Normal mode of operation
- Voice with visual display - Alexa has cards that show up in the app
What follows is a very abrivated version of what can be found in the [Alexa designing
for voice](https://developer.amazon.com/designing-for-voice/) documentation.
---
# Voice UX/UI - Design Process
Establish the purpose and user stories. The following are a couple parts of this process.
- What is the purpose of the skill? Why will people want to use it?
- Example: *To provide a way to learn a subject through flash cards*
- What can a user do with the skill?
- Example: *Ask for a random flash card*
---
# Voice UX/UI - Design Process
Write scripts
- Scripts for voice interfaces are like wireframes for a visual user interface
- Create dialogs between the user and the skill
```
User: Alexa, open flash card
Alexa: Welcome to flash card. Would you like a flash card?
User: Yes
Alexa: What is the atomic mass of oxygen?
User: Give me the answer
Alexa: 15.999
...
```
- It can be helpful to record dialogs with another person
---
# Voice UX/UI - Design Process
Develop the flow
- Outline the happy path flow first, this will be the shortest dialog to
get to a completion and then build on that with variations
- Keep in mind that a user may be able to start anywhere in a flow so you
will need to be able to fill in the blanks by reordering a dialog. For
example:
```
User: I want to fly to Chicago from DC
Alexa: When do you wish to travel?
...
```
or
```
User: I want to go to Chicago
Alexa: How do you want to get to Chicago?
User: I want to fly there from DC
Alexa: When do you wish to travel?
...
```
---
# Voice UX/UI - User Requests
1. Finish designing before you build
2. Identify intents, Use the built-in intents
3. Identify utterances
4. Identify slots
---
# Voice UX/UI - Alexa Responses
1. Be brief
2. Speak and write naturally
3. Use conversation markers
4. Add variety
5. Provide definitive choices
6. Handle problems
---
class: middle, center
# Questions?
---
# Example Skill - Basics
- Flash card skill
- User asks for a new card
- Skill responds with a random card
- User asks for the answer
- Skill responds with the answer to the card
We will need to hold on to some state after the skill responds with the
random card so we know what answer to give when asked.
---
# Example Skill - Heroku
We will need to get the sample deployed to Heroku and note the URL it gives
us for the deployment.
- Get a copy of the workshop repo:
```
git https://github.com/ornlabs/gwu_alexa_workshop
```
- Get the repo deployed to Heroku (note the URL that is returned by the push command)
```
npm install
heroku create
git push heroku master
```
---
# Example Skill - Setup
- Setup a skill in the Amazon Developer Console for Alexa.
- Skill Information
- Interaction builder - Intents
- **NextCardIntent**
- **AnswerCardIntent**
- Configuration
- Service endpoint selection should be HTTPS
- Set the "Default" field with your Heroku URL
- SSL Certificate
- Set "Certificate for DEFAULT Endpoint:" to the second option
- Test
---
# Example Skill - Publish
- Publishing Your Skills
- Skill Certification Process
---
# More resources
- Amazon guide: [Steps to build a custom skill](https://developer.amazon.com/docs/custom-skills/steps-to-build-a-custom-skill.html)
- Amazon guide: [Build skills with the alexa skills kit](https://developer.amazon.com/docs/ask-overviews/build-skills-with-the-alexa-skills-kit.html)
- [Codecademy Alexa Course](https://www.codecademy.com/learn/learn-alexa)
- [Alexa on github](https://github.com/alexa)
- [Alexa skills SDK for Javascript](https://github.com/alexa/alexa-skills-kit-sdk-for-nodejs)
- [Recent roundup of Alexa resources](https://developer.amazon.com/blogs/alexa/post/82578e06-1733-49ee-a801-e5bea45b2686/resource-roundup-top-alexa-tips-and-tutorials-for-new-skill-builders)
- [Recent roundup of Alexa tutorials](https://developer.amazon.com/blogs/alexa/post/762fee8e-9576-4c70-88c5-418ac809413f/resource-roundup-top-alexa-tips-and-tutorials-for-building-skills-with-visual-components)
- [Alexa videos from re:invent](https://developer.amazon.com/alexa-skills-kit/alexa-at-aws-reinvent)
- [Storyline](https://getstoryline.com/) - A newish visual builder for Alexa skills
- [Echosim](https://echosim.io/welcome) - An Echo simulator
</textarea>
<script src="https://remarkjs.com/downloads/remark-latest.min.js">
</script>
<script>
var slideshow = remark.create();
</script>
</body>
</html>