Voice User Interfaceの歴史と未来@Developers.IO 2017 WORLD 仙台

2017年11月11日に開催したクラスメソッド技術イベント「Developers.IO 2017 WORLD in 仙台」の発表資料です。

Transcript

1. History and future of Voice User Interface Ϋϥεϝιουגࣜձࣾ
   
   ϞόΠϧΞϓϦαʔϏε෦
   
   ਡ๚
   ༔ل %FWFMPQFST*0
   2017 WORLD SENDAI 
   2017.11.11
2. YUKI SUWA Technical Manager Classmethod, Inc.
3. ຊηογϣϯͷ಺༰ wਓྨ͕7PJDF
   6TFS
   *OUFSGBDFʹͨͲΓண͘·Ͱͷ
 ಓͷΓΛ୳ࡧ͠Α͏
   ‣Կͷҝʹੜ·ΕɺͲͷΑ͏ʹ੒௕͍ͯ͠Δͷ͔ʁ
   w7PJDF
   6TFS
   *OUFSGBDF͕໨ࢦ͍ͯ͠ΔະདྷΛ஌Ζ͏
   ‣ࢲͨͪιϑτ΢ΣΞɾϕϯμʔ͸
 7PJDF
   6TFS
   *OUFSGBDFͱ͍͏ٕज़Λ࢖ͬͯ
 Ͳ͏ͷΑ͏ͳ69ΛఏڙͰ͖Δఏڙ͢΂͖ͳͷ͔ʁ
4. Agenda w7PJDF
   6TFS
   *OUFSGBDFͱ͸
   w6TFS
   *OUFSGBDFͷྺ࢙
   w7PJDF
   6TFS
   *OUFSGBDFͷٕज़ͷྺ࢙
   w7PJDF
   6TFS
   *OUFSGBDFͷະདྷ
   w·ͱΊ 10 min 10 min 15 min 5 min 5 min
5. Voice User Interface (VUI) ͱ͸
6. Voice User Interfaceͱ͸ w੠ʹΑΔΠϯλʔ ϑΣʔ ε
   w࢖͍खͷ੠Ͱೖྗ͠ɺ࢖͍खʹ੠Ͱग़ྗ͢Δ
 ͍ͣΕ͔ͷΈͷ৔߹΋͋Δ
   wର࿩తͳ΍ΓͱΓͰίϯϐϡʔλΛૢ࡞͢Δ
7. ۙ͝ΖͷVoice User Interface
29. ࡢࠓͷ
    VUI ࣄ৘͸·͞ʹ
    େ
    PersonalAssistant ࣌୅
31. &BSCVET
35. 3PCPUT
39. ਓྨ͕76*ʹͨͲΓண͘·Ͱ
    History of User Interface
40. User Interfaceͱ͸ wਓؒͱίϯϐϡʔλؒͰ৘ใΛ΍ΓऔΓ͢Δ઀఺
41. CUI
    Character User Interface wจࣈྻʹΑΔ΍ΓऔΓΛߦ͏Πϯλʔ ϑΣʔ ε
    w͍ΘΏΔࠇ͍ը໘
    wΩʔϘʔυͳͲ͔ΒจࣈྻΛೖྗ͠ɺ
 ΢Πϯυ΢ʹจࣈྻΛग़ྗ͢Δ
    w$-*
    ͱ΋ݴ͏
43. CUIͷͭΒ͞ wجຊతʹ։ൃऀ͕࢖͏΋ͷ
    wҰൠϢʔβ޲͚Ͱ͸ͳ͍
    wίϚϯυΛ֮͑Δඞཁ͕͋Δ
    w௚ײతͰ͸ͳ͍
44. wίϯϐϡʔλɾάϥϑΟοΫεΛ༻͍ͯ΍ΓऔΓΛ ߦ͏Πϯλʔ ϑΣʔ ε
    w৘ใΛάϥϑΟΧϧʹදݱ
    wΑΓ௚ײతͳදݱʹ GUI
    Graphical User Interface
45. First GUI Xerox Star Apple Lisa Xerox Alto 1973 1981 1983
46. w%FTLUPQ
    .FUBQIPS
    w4LFVPNPSQIJTN
    %FTJHO
    w'MBU
    %FTJHO
    w.BUFSJBM
    %FTJHO History of GUI Design
52. wΑΓࣗવମͳೖग़ྗΛߦ͏Πϯλʔ ϑΣʔ ε
    wࣗવͳಈ࡞ͰίϯϐϡʔλΛૢ࡞͢Δ
    wը໘ʹ௚઀৮ΕͨΓɺ਎ମΛಈ͔͢
    wޒײʹಇ͖͔͚ΔϑΟʔυόοΫ NUI
    Natural User Interface
60. δΣενϟʔͷͭΒ͞ wख͕ർΕΔ
    w৔ॴΛऔΔ
    w͕͔͔࣌ؒΔ
    wޡૢ࡞͠΍͍͢
    wήʔϜ͸ָ͍͠
    wλεΫΛ͜ͳ͢ͷ͸ࠔ೉
61. NUI going to VUI w੠͸ը໘ʹґଘ͠ͳ͍
    w੠͸ࣗવʹൃ͢Δ͜ͱ͕Ͱ͖Δ
    wԻ͸ࣗવʹฉ͘͜ͱ͕Ͱ͖Δ
    w͍ͭͰ΋ɺͲ͜Ͱ΋࢖͑Δ
62. ίϯϐϡʔλΛ࢖͍ͬͯΔ
    ͱ͍͏ײ͕֮ബΕ͖͍ͯͯΔ
67. VUIΛऔΓרٕ͘ज़ͷྺ࢙
68. VUIͷΞʔΩςΫνϟ https://lekta.ai/blog/talking-to-machines-more-naturally-than-ever-before-voice- interface-for-lekta-nlp/ ԿΛೖྗͤ͞Δ͔ʁ Ͳ͏൓Ԡ͢Δ͔ʁ
69. VUIͷΞʔΩςΫνϟ ԿΛೖྗͤ͞Δ͔ʁ
70. History of Speech Recognition 1952 Audery by Bell Labs 1962 Shoebox by IBM 1970 Hidden Markov Model 1971 SUR Program by DARPA 1980s IVR by Speechworks 1990s Dragon Dictate 2000s Siri by Siri Voice Search by Google ਺ࣈ
    
    
    
    ͷԻ੠ೝࣝ ୯ޠͷԻ੠ೝࣝ ୯ޠͷԻ੠ೝࣝ Ի੠ೝࣝͷੑೳͷ޲্ IVRͷىݯ ࿈ଓͨ͠Ի੠ೝࣝ Personal Assistantͷىݯ by Nuance Communications
71. Shoebox
    IBM 1962
72. Hidden Markov Model wͦΕ·ͰͷԻ੠ೝٕࣝज़ͷ՝୊Λղܾ͢Δํ๏
    w౷ܭσʔλΛݩʹ֬཰తʹσʔλΛѻ͏
    w࣌ؒมԽ΋ؚΊͯԻ੠Λ౷ܭతʹදݱͰ͖Δ http://www.tel.co.jp/museum/magazine/human/120709_topics_03/index.html
73. IVR
    Interactive Voice Response
74. Twilio
75. Amazon Connect
76. Siri
    Siri 2007
77. Siri … ?
78. Speech Interpretation and Recognition Interface Ի੠ղऍͱԻ੠ೝࣝͷ ΠϯλʔϑΣʔε
79. Voice Search
    Google 2008
80. AI & Chat Bot
81. VUIͷΞʔΩςΫνϟ Ͳ͏൓Ԡ͢Δ͔ʁ
82. History of Chat Bot 1966 ELIZA 1972 PARRY 1988 JABBERWACKY 1995 A.L.I.C.E. 2010 Siri 2012 Google Now ੈքॳͷࣗಈԠ౴ ҙݟΛ࣋ͭ&-*;" ݴޠͱจ຺͔ΒԠ౴Λ࡞੒ ΦʔϓϯιʔεԽ εϚϗͰಈ࡞͢ΔBot
83. ػց࿦
    René Descartes ఱީ΍஍ܗɺੜ໋ͳͲͷॾʑ ͷࣗવݱ৅Λͦͷ෦෼͍҃ ͸ཁૉͷܾఆ࿦తͳҼՌؔ ܎ͷΈɺಛʹݹయྗֶతͳ ҼՌ࿈࠯ͷΈͰɺղऍͰ͖ શମͷৼΔ෣͍΋༧ଌͰ͖ Δͱ͢Δཱ৔
84. ELIZA wδϣηϑɾϫΠθϯό΢ϜʹΑͬͯൃ໌
    wདྷஊऀத৺ྍ๏Λߦ͏ͨΊʹ࡞ΒΕͨ
    w؆୯ͳύλʔϯϚονϯάٕ๏ʹΑΔࣗવݴޠॲཧ
87. ਓʹدΓఴ͏Α͏ʹ
    ਐԽ͖ͯͨ͠
88. VUIͷະདྷ
89. ੜ׆ͷதͰࣗવʹ࢖͑ΔΑ͏ʹ wΑΓࣗવମͳձ࿩
    wΑΓεϚʔτͳࢧԉ
    wΑΓଟ͘ͷ෼໺Ͱͷ׆༻
90. ΑΓࣗવମͳձ࿩ wձ࿩ͷίϯςΩετΛཧղ͢Δ
    w࿩ͷྲྀΕΛཧղ͢Δ
    wର࿩૬खʢϢʔβʔʣͷ͜ͱΛཧղ͢Δ
    wঢ়گΛཧղ͢Δ
91. ΑΓεϚʔτͳࢧԉ w༧ଌม׵
    w$6*
    ΍
    (6*
    ͱͷ૊Έ߹Θͤ
    wϋʔυ΢ΣΞ΋ؚΊͨ૯߹తͳମݧ
92. ΑΓଟ͘ͷ෼໺Ͱͷ׆༻
93. VUI
    ͔Β͞ΒͳΔະདྷ΁
94. wঢ়گʹ߹ΘͤͯมԽ͢ΔΠϯλʔ ϑΣʔ ε
    wػೳͱܗଶ͕ҰମԽ͢Δ
    wίϯϐϡʔλ͕ΑΓੜ׆ʹ༹͚ࠐΜͰ͍͘ OUI
    Organic User Interface
95. IUUQHJIZPKQEFTJHOTFSJBMXTFB ී௨ɺਓؒͬͯѲख͢Δͱɺඵ͘Β͍Ͱ΋ɺྗͷ ೖΕ۩߹ͱ͔ৼΓ۩߹ͱ͔Ͱɺ૬ख͕ࣗ෼ΛͲ͏ࢥͬ ͍ͯΔ͔Θ͔ͬͪΌ͏ΜͰ͢Ͷɺ͍͋͋͏ײ৮ͱ͍͏ ͷ͸ɺϚ΢εͰΫϦοΫͯ͠ϝχϡʔ͕ग़ΔΑΓΉ͠ Ζૣ͍ɻ͜͏͍͏ΠϯλʔϑΣʔε͸ࠓͷٕज़ͩͬͨ Βे෼Ͱ͖ΔΜ͡Όͳ͍͔ͱɻ l
103. https://youtu.be/JOaGTIVyS6U
104. ·ͱΊ w͍ͭͷ࣌୅΋ੜ׆Λ๛͔ʹ͢Δ͜ͱ͕໨త
     wίϯϐϡʔλ͕ɺΑΓੜ׆ʹ༹͚ࠐΜͰ͍͘
     wਓؒΛஔ͖׵͑Δଘࡏʹ
     w76*Λ׆༻ͯ͠ͲͷΑ͏ʹੜ׆Λ๛͔ʹͰ͖Δ͔
 Λߟ͑ͯ࡞Ζ͏