Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

(MBL310) Alexa Voice Service Under the Hood

Alexa is the speech processing and personal assistant technology behind Amazon Echo. Speech-based user interfaces represent one of the next major disruptions in computing and the Alexa Voice Service (AVS) provides you with an opportunity to take advantage of this new form of interaction.  In this session, we’ll walk through the recently-released AVS API by building a voice-enabled application and then go behind the scenes with Alexa, diving into the architecture and unique technical challenges faced during development.

  • Soyez le premier à commenter

(MBL310) Alexa Voice Service Under the Hood

  1. 1. © 2015, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Donn Morrill - Sr. Manager, Solutions Architecture, Amazon Alexa Frederic Deramat - Sr. Principal Engineer, Amazon Alexa Stephen Chien – Senior Advanced Technology Engineer, BMW October 2015 MBL310 Alexa Voice Service Under the Hood
  2. 2. Agenda • Echo demo • What is AVS? • AVS walkthrough • AVS tips and tricks • Alexa under the hood • Alexa in action - BMW
  3. 3. Echo demo
  4. 4. What is ASK? The Alexa Skills Kit (ASK) is a collection of self-service APIs, tools, documentation, and code samples that make it fast and easy for you to add skills to Alexa. Using ASK, you can teach new skills to Alexa in just a few hours.
  5. 5. Alexa companion app • Register a device • View and manage actions • Link third-party accounts • View lists • And much more
  6. 6. What is AVS?
  7. 7. Registration and authentication
  8. 8. Basic AVS request 1) Create an HTTP request 2) Collect the audio and send the request • Use “Transfer-Encoding: Chunked” for streaming the request or set “Content-Length” for sending prerecorded audio • Avoid storing the audio 3) Parse the response 4) Handle each directive
  9. 9. Basic AVS request: Write HTTP headers POST /v1/avs/speechrecognizer/recognize HTTP/1.1 Host: access-alexa-na.amazon.com Authorization: Bearer {ACCESS TOKEN}  From auth flow Content-Type: multipart/form-data; boundary={BOUNDARY TERM} Transfer-Encoding: chunked OR Content-Length: {LENGTH}
  10. 10. Basic AVS request: Send the audio --{BOUNDARY TERM} Content-Disposition: form-data; name="request" Content-Type: application/json; charset=UTF-8 "messageBody": { "profile": "alexa-close-talk",  Speech model "locale": "en-us", "format": "audio/L16; rate=16000; channels=1" } --{BOUNDARY TERM} Content-Disposition: form-data; name="audio" Content-Type: audio/L16; rate=16000; channels=1 {ENCODED AUDIO}  16 bit linear PCM at 16 kHz
  11. 11. Basic AVS request: Parse the response HTTP/1.1 200 OK Content-Type: multipart/related; boundary={BOUNDARY TERM} --{BOUNDARY TERM} Content-Type: application/json; charset=UTF-8 { "messageHeader": {}, "messageBody": { "directives": [ { "namespace": “{AudioPlayer | SpeechSynthesizer}", "name": “{DIRECTIVE}", "payload" : { --{BOUNDARY TERM} Content-Disposition: form-data; name="audio" Content-Type: audio/mpeg Content-ID: {CONTENT ID} {ENCODED AUDIO DATA}  If directive is speak, play this back to the user
  12. 12. AVS response directives Directive Meaning clearQueue Clear the playback queue listen Multi-turn interaction - listen for more audio play Play a streaming URL speak Speak the Alexa response to the user stop Stop streaming playback
  13. 13. Streaming events Event Meaning metadataAvailable Metadata associated with the selection in the audio player playbackError Audio player has experienced an issue during playback playbackFinished Audio player has completed playback playbackIdle Audio player has reached the end of the playlist playbackInterrupted Audio player has been interrupted playbackProgressReport Current state of the audio player playbackStarted Audio player has started playing
  14. 14. Alexa platform
  15. 15. AVS interfaces Interface Directives Events SpeechRecognizer onSpeech onTimeout SpeechSynthesizer speak ask AudioPlayer play stop onPlaybackStarted onPlaybackFinished onPlaybackNearlyFinished Speaker setVolume onVolumeChange
  16. 16. SpeechRecognizer
  17. 17. Connected Home
  18. 18. AudioPlayer
  19. 19. One-to-many interfaces
  20. 20. Close-talk vs. far-field Protocol Client Components HTTP REST HTTP REST Bidirectional async Endpointer Communications Communications Beamforming Echo cancellation Wakeword Bidirectional async 1 2 3 4
  21. 21. Endpoint and protocol independence
  22. 22. AVS best practices • Stream • Decouple up and down channels • Decouple protocol • Adopt event-driven patterns
  23. 23. AVS under the hood
  24. 24. Under the hood: SLU
  25. 25. Skill orchestration
  26. 26. Skill orchestration: Focus & Context
  27. 27. Skill orchestration
  28. 28. Skill orchestration: Connected Home
  29. 29. Continuous improvement: Offline
  30. 30. Continuous improvement: Online
  31. 31. Continuous improvement: Online
  32. 32. Continuous improvement: Online
  33. 33. SMART AGENTS FOR BMW DRIVERS. BMW Group Technology Office USA. Stephen Chien
  34. 34. THE AUTOMOTIVE INDUSTRY WILL UNDERGO MASSIVE CHANGES IN THE NEXT 10 YEARS. SUSTAINABILITY DIGITALIZATION CARS ARE POWERED BY ELECTRICITY CO2 SUSTAINABILITY HAS GAINED GREATER SIGNIFICANCE VEHICLES ARE DRIVING ALMOST ACCIDENT-FREE MOBILITY CONCEPTS LIKE CAR- SHARING ARE WIDELY AVAILABLE VEHICLES BECOME PART OF THE DIGITAL ECOSYSTEM HUMAN-MACHINE INTERFACES BECOME MORE INTUITIVE
  35. 35. BMW i – THE NEXT GENERATION OF A CONNECTED CAR. Wearable Integration Smart TV Integration Smartphone Integration
  36. 36. PERSONAL SMART AGENTS FOR BMW DRIVERS. • Smart agents are everywhere in life. • Natural way to interact with your vehicle. • Easy for users to understand and use. • A smart agent that knows about you and your vehicle.
  37. 37. PROTOTYPING AN ALEXA SKILL FOR BMW i. Leveraging Amazon’s developer platform. • Built on AWS technology. • ASK and AVS provide a seamless user experience. • ASK and AVS provide open platforms. BMW skill features. • Request vehicle status through Alexa. • Execute vehicle commands through Alexa.
  38. 38. BMW i SKILL DEMO.
  39. 39. Thank you! Drawing winners announced: http://bit.ly/appstoregiveaway
  40. 40. Remember to complete your evaluations!

    Soyez le premier à commenter

    Identifiez-vous pour voir les commentaires

  • AdetunjiAdegbesan

    Mar. 30, 2016
  • sheshech

    Apr. 23, 2016
  • SadanandMandal

    Jul. 22, 2016
  • vhsu

    Oct. 19, 2016
  • GwendolynErvinPMPCSP

    Oct. 26, 2016
  • batpig61

    Nov. 22, 2016
  • JorgeRSantos

    Dec. 31, 2016
  • duckjaechae

    Feb. 13, 2017
  • webispy

    Mar. 21, 2017
  • ShinsukeYokota1

    Apr. 22, 2017
  • GuidoDAlbore

    May. 24, 2017
  • alexcee

    Aug. 8, 2017
  • ShinsukeYokota

    Nov. 27, 2017
  • JungminAn

    Apr. 23, 2018
  • ShunFangYang

    Jul. 27, 2018
  • Davidtung888

    Nov. 20, 2018
  • javadance

    Feb. 5, 2019

Alexa is the speech processing and personal assistant technology behind Amazon Echo. Speech-based user interfaces represent one of the next major disruptions in computing and the Alexa Voice Service (AVS) provides you with an opportunity to take advantage of this new form of interaction.  In this session, we’ll walk through the recently-released AVS API by building a voice-enabled application and then go behind the scenes with Alexa, diving into the architecture and unique technical challenges faced during development.

Vues

Nombre de vues

9 375

Sur Slideshare

0

À partir des intégrations

0

Nombre d'intégrations

26

Actions

Téléchargements

619

Partages

0

Commentaires

0

Mentions J'aime

17

×