AWS CDK で文字起こし + 翻訳パイプラインを作ってみた /jawsug-bgnr-26-lt

JAWS-UG 初心者支部 #26 の LT スライドです。#24 で宿題となっていた Amazon Transcribe と Amazon Translate、および Amazon S3 と AWS Lambda を用いた文字起こし + 翻訳パイプラインの作成を、 AWS CDK でやってみた という内容です。

Transcript

1. 
   ҏ౻Յ༸
   
   :PTIJIJSP
   *UP "84
   $%,
   Ͱ
   จࣈى͜͠
   
   ຋༁ύΠϓϥΠϯΛ
   ࡞ͬͯΈͨ <+"846(
   ॳ৺ऀࢧ෦
   > KBXTVH@CHOS
   KBXTVHDIJCB
2. 
   
   ॴଐɾ୲౰
   ɹɹ
   גࣜձࣾϞϦαϫ
   ɹɹ
   ΠϯϑϥɾαʔόαΠυ
   
   
   ޷͖ͳ
   "84
   ͷαʔϏε
   ɹɹ
   "84
   -BNCEB
   ɹɹ
   "84
   $FSUJGJDBUF
   .BOBHFS
   
   
   झຯ
   ɹɹ
   εϐʔυΩϡʔϒ
   ɹɹ
   ಓͷӺΊ͙Γ
   
   
   ϒϩά
   ɹɹ
   IUUQTNJDIJNBOJOFU ࣗݾ঺հ ҏ౻
   Յ༸
   
   :PTIJIJSP
   *UP
   ɹɹ!NJDIJNBOJ
3. 
   ॓୊ͷ಺༰Λ͓͞Β͍
   
   "84
   $%,
   ͱ͸
   
   "84
   $%,
   Ͱ॓୊΍ͬͯΈͨ
   
   ·ͱΊ
   ˞࿩͞ͳ͍͜ͱ
   
   
   "84
   $%,
   ͷৄࡉͳઆ໌
   
   
   -BNCEB
   ؔ਺ͷ࣮૷಺༰ͷৄࡉ ΞδΣϯμ
4. 
   ॓୊ͷ಺༰Λ͓͞Β͍
5. 
   ॓୊ͷ಺༰Λ͓͞Β͍
6. 
   ॓୊ͷ಺༰Λ͓͞Β͍
7. 
   "84
   $%,
   ͱ͸ IUUQTEPDTBXTBNB[PODPNDELMBUFTUHVJEFIPNFIUNM
8. 
   "84
   $%,
   ͱ͸ IUUQTEPDTBXTBNB[PODPNDELMBUFTUHVJEFIPNFIUNM ͬ͘͟Γ຋༁͢Δͱɺ"84
   $%,
   ͸
   $MPVE'PSNBUJPO
   Λհͯ͠
   "84
   ͷϦιʔεΛ؅ཧ͢ΔͨΊͷϑϨʔϜϫʔΫ
9. 
   "84
   $%,
   ͱ͸ ͭ·ΓɺͲ͏͍͏͜ͱʁ
10. 
    "84
    $%,
    ͱ͸ ͭ·ΓɺͲ͏͍͏͜ͱʁ
     "84
    ͷϦιʔεΛओཁͳݴޠͰɺϓϩάϥϜΛॻ͘Α͏ʹఆٛͰ͖Δ
    5ZQF4DSJQU
    1ZUIPO
    ͳͲ
     ϓϩάϥϜΛ΋ͱʹͯ͠
    $MPVE'PSNBUJPO
    ςϯϓϨʔτΛੜ੒Ͱ͖Δ
     DEL
    ίϚϯυͰςϯϓϨʔτͷ࣮ߦ·ͰՄೳ
    
11. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ
12. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ -BNCEB
    ؔ਺ͷ࡞੒
     จࣈى͜͠ॲཧ༻
    
    transcribe.py ϋϯζΦϯΛ΄΅ͦͷ··
     ຋༁ॲཧ༻
    
    translate_from_transcribe.py
13. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ ֤Ϧιʔεͷఆٛ
     4
    όέοτ
    Y
     -BNCEB
    ؔ਺
    Y
     *".
    ϩʔϧɺ
    *".
    ϙϦγʔ
     4
    Πϕϯτιʔε
    ˞
    $%,
    ͷϓϩάϥϜ͸
    5ZQF4DSJQU
    Ͱॻ͍͍͖ͯ·͢
14. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ ֤Ϧιʔεͷఆٛ
     4
    όέοτ
    Y
     -BNCEB
    ؔ਺
    Y
     *".
    ϩʔϧɺ
    *".
    ϙϦγʔ
     4
    Πϕϯτιʔε
    ˞
    $%,
    ͷϓϩάϥϜ͸
    5ZQF4DSJQU
    Ͱॻ͍͍͖ͯ·͢
15. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ 4
    όέοτͷఆٛ
    
16. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ 4
    όέοτͷఆٛ
    ͦΕͧΕͷ
    4
    όέοτ໊Λ
    ഑ྻͰఆٛ
17. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ 4
    όέοτͷఆٛ
    ͦΕͧΕͷ
    4
    όέοτ໊Λ
    ഑ྻͰఆٛ GPS&BDI
    Ͱ
    
    4
    όέοτΛఆٛ
18. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ -BNCEB
    ؔ਺ͷఆٛ
    
19. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ -BNCEB
    ؔ਺ͷఆٛ
    -BNCEB
    ؔ਺
    ຊମͷఆٛ
20. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ -BNCEB
    ؔ਺ͷఆٛ
    *".
    ϩʔϧΛ௥Ճ -BNCEB
    ؔ਺
    ຊମͷఆٛ
21. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ -BNCEB
    ؔ਺ͷఆٛ
    *".
    ϩʔϧΛ௥Ճ ΠϕϯτιʔεΛ௥Ճ -BNCEB
    ؔ਺
    ຊମͷఆٛ
22. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ $MPVE'PSNBUJPO
    ςϯϓϨʔτੜ੒
    
23. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ $MPVE'PSNBUJPO
    ςϯϓϨʔτੜ੒
    
24. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ $MPVE'PSNBUJPO
    ςϯϓϨʔτੜ੒
    
25. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ σϓϩΠ
    
26. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ σϓϩΠ
    
27. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ σϓϩΠ
    
28. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ σϓϩΠ
    
29. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ σϓϩΠ
    
30. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ σϓϩΠ
    111111111111
31. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ ಈ࡞֬ೝ
32. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ ಈ࡞֬ೝ )FMMP
    %P
    ZPV
    TQFBL
    B
    GPSFJHO
    MBOHVBHF
    0OF
    MBOHVBHF
    JT
    OFWFS
    FOPVHI
33. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ ಈ࡞֬ೝ )FMMP
    %P
    ZPV
    TQFBL
    B
    GPSFJHO
    MBOHVBHF
    0OF
    MBOHVBHF
    JT
    OFWFS
    FOPVHI ❯ aws s3 cp sample.mp3 s3://transcribe-input-bucket upload: sample.mp3 to s3://transcribe-input-bucket/sample.mp3
34. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ ಈ࡞֬ೝ )FMMP
    %P
    ZPV
    TQFBL
    B
    GPSFJHO
    MBOHVBHF
    0OF
    MBOHVBHF
    JT
    OFWFS
    FOPVHI ❯ aws s3 cp sample.mp3 s3://transcribe-input-bucket upload: sample.mp3 to s3://transcribe-input-bucket/sample.mp3 ͠͹Β͘͢Δͱʜ
35. 
    "84
    $%,
    Ͱ॓୊΍ͬͯΈͨ ಈ࡞֬ೝ )FMMP
    %P
    ZPV
    TQFBL
    B
    GPSFJHO
    MBOHVBHF
    0OF
    MBOHVBHF
    JT
    OFWFS
    FOPVHI ❯ aws s3 cp sample.mp3 s3://transcribe-input-bucket upload: sample.mp3 to s3://transcribe-input-bucket/sample.mp3 ͠͹Β͘͢Δͱʜ
36. 
    ·ͱΊ  "NB[PO
    4
    ͸ͨͩͷετϨʔδͰ͸ͳ͍
     αʔόϨεָ͍͠
     "84
    $%,
    ָ͍͠
    ˙
    ϒϩάهࣄ
    +"846(
    ॳ৺ऀࢧ෦
    αʔόϨεϋϯζΦϯษڧձͷ॓୊Λ΍ͬͯΈͨ
    KBXTVH@CHOS
    KBXTVH
    IUUQTNJDIJNBOJOFUQPTUBXTKBXTVHCHOSTFSWFSMFTTRVJLDIBOETPO
    ˙
    $%,
    ϓϩδΣΫτͷιʔείʔυ
    IUUQTHJUIVCDPNNJDIJNBOJKBXTVHCHOSIPNFXPSL
37. ͓͠·͍ KBXTVH@CHOS
    KBXTVHDIJCB