I-Amazon Transcribe iyisevisi ye-AWS evumela amakhasimende ukuthi aguqule inkulumo ibe umbhalo ngeqoqo noma imodi yokusakaza. Isebenzisa ukuqashelwa kwenkulumo okuzenzakalelayo okunikwe amandla okufunda (ASR), ukuhlonza ulimi okuzenzakalelayo, nobuchwepheshe bokucubungula ngemva kokucubungula. I-Amazon Transcribe ingasetshenziselwa ukuloba izingcingo zokunakekelwa kwamakhasimende, izingcingo zezingqungquthela zamaqembu amaningi, nemilayezo yevoyisimeyili, kanye nokukhiqizwa kwemibhalo engezansi yamavidiyo arekhodiwe nabukhoma, ukusho izibonelo ezimbalwa nje. Kulokhu okuthunyelwe kwebhulogi, uzofunda ukuthi unganika kanjani amandla izinhlelo zakho zokusebenza ngamakhono we-Amazon Transcribe ngendlela ehlangabezana nezidingo zakho zokuphepha.

Amanye amakhasimende aphathisa i-Amazon Transcribe idatha eyimfihlo futhi ephathelene nebhizinisi labo. Kwezinye izimo, okuqukethwe komsindo okucutshungulwe yi-Amazon Transcribe kungase kuqukathe idatha ebucayi okudingeka ivikelwe ukuze ihambisane nemithetho yendawo nezimiso. Izibonelo zalolo lwazi ulwazi oluhlonza umuntu (PII), ulwazi lwezempilo lomuntu siqu (PHI), kanye nedatha yemboni yekhadi lokukhokha (PCI). Ezigabeni ezilandelayo zebhulogi, sihlanganisa izindlela ezihlukene i-Amazon Transcribe enazo ukuze ivikele idatha yekhasimende ekuthuthweni nasekuphumuleni. Sabelana ngalezi zindlela ezihamba phambili zokuphepha eziyisikhombisa zokwakha izinhlelo zokusebenza nge-Amazon Transcribe ehlangabezana nezidingo zakho zokuphepha nokuhambisana:

1. Sebenzisa ukuvikela idatha nge-Amazon Transcribe
2. Xhumana ngendlela yenethiwekhi eyimfihlo
3. Lungisa idatha ebucayi uma kudingeka
4. Sebenzisa izindima ze-IAM ezinhlelweni zokusebenza namasevisi e-AWS adinga ukufinyelela kwe-Amazon Transcribe
5. Sebenzisa ukulawula kokufinyelela okusekelwe kumaka
6. Sebenzisa amathuluzi wokuqapha we-AWS
7. Nika amandla i-AWS Config

Izinqubo ezihamba phambili ezilandelayo ziyiziqondiso ezivamile futhi azimeli isixazululo esiphelele sokuvikela. Ngenxa yokuthi lezi zinqubo ezihamba phambili zingase zingafaneleki noma zanele indawo okuyo, zisebenzise njengokucatshangelwa okuwusizo kunemiyalelo.

Umkhuba ongcono kakhulu 1 - Sebenzisa ukuvikela idatha nge-Amazon Transcribe

I-Amazon Transcribe ihambisana ne Imodeli yemfanelo eyabiwe ye-AWS, ehlukanisa isibopho se-AWS sokuphepha kwefu kusukela kumthwalo wekhasimende ngokuvikeleka emafini.

I-AWS inesibopho sokuvikela ingqalasizinda yomhlaba wonke esebenzisa yonke i-AWS Cloud. Njengekhasimende, unesibopho sokugcina ukulawula okuqukethwe kwakho okusingathwa kule ngqalasizinda. Lokhu okuqukethwe kufaka phakathi ukulungiselelwa kokuvikeleka nemisebenzi yokuphatha yamasevisi e-AWS owasebenzisayo. Ukuze uthole ulwazi olwengeziwe mayelana nobumfihlo bedatha, bona I-FAQ Eyimfihlo Yedatha.

Ivikela idatha kwezokuthutha

Ukubethelwa kwedatha kusetshenziselwa ukwenza isiqiniseko sokuthi ukuxhumana kwedatha phakathi kwesicelo sakho ne-Amazon Transcribe kuhlala kuyimfihlo. Ukusetshenziswa kwe-cryptographic algorithms eqinile kuvikela idatha ngenkathi idluliswa.

I-Amazon Transcribe ingasebenza ngenye yalezi zindlela ezimbili:

• Ukusakaza okulotshiweyo vumela ukuloba kokusakaza kwemidiya ngesikhathi sangempela
• Imisebenzi yokulotshwa kweqoqo vumela ukulotshwa kwamafayela alalelwayo kusetshenziswa imisebenzi engavumelanisi.

Kumodi yokusakaza yokuloba, izinhlelo zokusebenza zeklayenti zivula uxhumo lokusakaza-bukhoma oluqondiswe kabili nge-HTTP/2 noma i-WebSockets. Uhlelo lokusebenza luthumela ukusakazwa komsindo ku-Amazon Transcribe, futhi isevisi iphendula ngokusakaza kombhalo ngesikhathi sangempela. Kokubili ukuxhumana kokusakaza kwe-HTTP/2 kanye ne-WebSockets kusungulwa nge-Transport Layer Security (TLS), okuyiphrothokholi eyimfihlo eyamukelwa kabanzi. I-TLS ihlinzeka ngokufakazela ubuqiniso nokubethela kwedatha kwezokuthutha kusetshenziswa izitifiketi ze-AWS. Sincoma ukusebenzisa i-TLS 1.2 noma kamuva.

Kumodi ye-batch transcription, ifayela lomsindo kuqala lidinga ukufakwa ku- I-Amazon Simple Storage Service (I-Amazon S3) ibhakede. Bese umsebenzi wokulotshwa kwenqwaba okhomba i-S3 URI yaleli fayela uyadalwa ku-Amazon Transcribe. Kokubili i-Amazon Transcribe kumodi ye-batch kanye ne-Amazon S3 isebenzisa i-HTTP/1.1 phezu kwe-TLS ukuvikela idatha ekuthuthweni.

Zonke izicelo ku-Amazon Transcribe nge-HTTP namaWebSockets kumele zigunyazwe kusetshenziswa Inguqulo yesi-4 yesiginesha ye-AWS. Kunconywa ukusebenzisa iSiginesha Yenguqulo 4 ukuze uqinisekise izicelo ze-HTTP ku-Amazon S3 futhi, nakuba ukuqinisekiswa kwabadala. Inguqulo yesi-2 yesiginesha kungenzeka nakwezinye izifunda ze-AWS. Izicelo kufanele zibe nemininingwane evumelekile ukuze zisayine izicelo ze-API kumasevisi e-AWS.

Ukuvikela idatha ngesikhathi sokuphumula

I-Amazon Transcribe kumodi ye-batch isebenzisa amabhakede e-S3 ukugcina kokubili ifayela lomsindo elifakiwe kanye nefayela eliphumayo lokuloba. Amakhasimende asebenzisa ibhakede le-S3 ukuze agcine ifayela lomsindo elifakiwe, futhi kunconywa kakhulu ukunika amandla ukubethela kuleli bhakede. I-Amazon Transcribe isekela izindlela ezilandelayo zokubethela ze-S3:

Zombili izindlela zibethela idatha yekhasimende njengoba ibhalwe kumadiski futhi isuse ukubhala phansi lapho ufinyelela kuyo usebenzisa eyodwa yama-block cyphers anamandla atholakalayo: 256-bit Advanced Encryption Standard (AES-256) GCM.Uma usebenzisa i-SSE-S3, okhiye bokubethela baphathwa futhi ijikeleziswa njalo yisevisi ye-Amazon S3. Ukuze uthole ukuvikeleka okwengeziwe nokuthobelana, i-SSE-KMS inikeza amakhasimende ukulawula kokhiye bokubethela ngokusebenzisa Isevisi yokuphatha ukhiye we-AWS (AWS KMS). I-AWS KMS inikeza izilawuli zokufinyelela ezengeziwe ngenxa yokuthi kufanele ube nezimvume zokusebenzisa okhiye abafanelekile be-KMS ukuze ubethele futhi uguqule ukubethela izinto kumabhakede e-S3 alungiselelwe nge-SSE-KMS. Futhi, i-SSE-KMS inikeza amakhasimende amandla okulandelela okucwaningwayo agcina amarekhodi okuthi ubani osebenzise okhiye bakho be-KMS nokuthi nini.

Okulotshiweyo kokuphumayo kungagcinwa kubhakede elifanayo le-S3 eliphethwe yikhasimende. Kulokhu, izinketho zokubethela ezifanayo ze-SSE-S3 ne-SSE-KMS ziyasebenza. Enye inketho yokuphuma kwe-Amazon Transcribe kumodi ye-batch isebenzisa ibhakede le-S3 elilawulwa yisevisi. Bese idatha yokuphumayo ifakwa kubhakede elivikelekile le-S3 eliphethwe isevisi ye-Amazon Transcribe, futhi unikezwa i-URI yesikhashana engasetshenziswa ukulanda okulotshiweyo kwakho.

I-Amazon Transcribe isebenzisa okubethelwe Isitolo se-Amazon Elastic Block (Amazon EBS) amavolumu okugcina idatha yekhasimende okwesikhashana phakathi nokucubungula imidiya. Idatha yekhasimende iyahlanzwa kuwo womabili amacala aphelele nawokuhluleka.

Umkhuba ongcono kakhulu 2 - Xhumana ngendlela yenethiwekhi yangasese

Amakhasimende amaningi athembele ekubetheleni kwezokuthutha ukuze axhumane ngokuphephile ne-Amazon Transcribe nge-inthanethi. Nokho, kwezinye izinhlelo zokusebenza, ukubethela kwedatha kukuthutha kungase kunganeli ukuhlangabezana nezimfuneko zokuphepha. Kwezinye izimo, idatha iyadingeka ukuze ungadluli kumanethiwekhi omphakathi njenge-inthanethi. Futhi, kungase kube nemfuneko yokuthi isicelo sisetshenziswe endaweni eyimfihlo engaxhumekile ku-inthanethi. Ukuze uhlangabezane nalezi zidingo, sebenzisa interface VPC endpoints Inikwe amandla yi- I-AWS PrivateLink.

Umdwebo olandelayo wezakhiwo ubonisa isimo sokusetshenziswa lapho isicelo senziwa khona I-Amazon EC2. Isenzakalo se-EC2 esisebenzisa uhlelo lokusebenza asinakho ukufinyelela ku-inthanethi futhi sixhumana ne-Amazon Transcribe kanye ne-Amazon S3 nge-interface ye-VPC endpoints.

Kwezinye izimo, uhlelo lokusebenza oluxhumana ne-Amazon Transcribe lungase lusetshenziswe esikhungweni sedatha esisendaweni. Kungase kube nezimfuneko ezengeziwe zokuphepha noma zokuthobelana ezigunyaza ukuthi idatha eshintshiwe ne-Amazon Transcribe akumele idlulise amanethiwekhi omphakathi njenge-inthanethi. Kulokhu, ukuxhumana okuyimfihlo nge I-AWS Direct Connect ingasetshenziswa. Umdwebo olandelayo ubonisa isakhiwo esivumela uhlelo lokusebenza olusemagcekeni ukuthi luxhumane ne-Amazon Transcribe ngaphandle kokuxhuma ku-inthanethi.

Indlela engcono kakhulu yesi-3 - Lungisa idatha ebucayi uma kudingeka

Ezinye izimo zokusetshenziswa nezimo zokulawula zingadinga ukususwa kwedatha ebucayi emibhalweni ebhaliwe namafayela alalelwayo. I-Amazon Transcribe isekela ukuhlonza kanye nokwenza kabusha imininingwane ehlonza umuntu (PII) njengamagama, amakheli, izinombolo zokuphepha komphakathi, nokunye. Leli khono lingasetshenziswa ukuze amakhasimende afinyelele ukuthobela imboni yekhadi lokukhokha (PCI) ngokukhipha kabusha i-PII efana nenombolo yekhadi lesikweletu noma lasebhange, usuku lokuphelelwa yisikhathi, kanye nekhodi yokuqinisekisa yekhadi elinezinombolo ezintathu (CVV). Imibhalo elotshiwe enolwazi oludakiwe izoba ne-PII esikhundleni sezimeli kubakaki abayisikwele ebonisa ukuthi hlobo luni lwe-PII olwenziwe kabusha. Okulotshiweyo kokusakaza-bukhoma kusekela ikhono elengeziwe lokuhlonza i-PII kuphela nokuyilebula ngaphandle kokuphinda kwenziwe kabusha. Izinhlobo ze-PII ezenziwe kabusha yi-Amazon Transcribe ziyahlukahluka phakathi kwenqwaba nokulotshiweyo kokusakaza. Bukela ku Ilungisa kabusha i-PII emsebenzini wakho weqoqo futhi Ukwenza kabusha noma ukuhlonza i-PII ngokusakaza kwesikhathi sangempela Ukuze uthole imininingwane eyengeziwe.

The ezikhethekile I-Amazon Transcribe Call Analytics Ama-API anekhono elakhelwe ngaphakathi lokuhlela kabusha i-PII kukho kokubili okulotshiweyo kombhalo namafayela alalelwayo. Le API isebenzisa amamodeli akhethekile enkulumo-kuya-umbhalo kanye nokucubungula ulimi lwemvelo (NLP) aqeqeshelwe ukuqonda isevisi yamakhasimende nezingcingo zokuthengisa. Kwezinye izimo, ungasebenzisa lesi sixazululo ukulungisa i-PII kumafayela alalelwayo nge-Amazon Transcribe.

Izindlela ezihamba phambili zokuphepha ze-Amazon Transcribe

Ukwenza okuhle kakhulu 4 - Sebenzisa Izindima ze-IAM ngezinhlelo zokusebenza nezinsiza ze-AWS ezidinga ukufinyelela kwe-Amazon Transcribe. Uma usebenzisa indima, akudingekile ukuthi usabalalise imininingwane yesikhathi eside, njengamaphasiwedi noma okhiye bokufinyelela, kusenzakalo se-EC2 noma isevisi ye-AWS. Izindima ze-IAM zinganikeza izimvume zesikhashana izinhlelo zokusebenza ezingazisebenzisa uma zenza izicelo kuzinsiza ze-AWS.

Ukuzijwayeza Okungcono Kakhulu 5 - Sebenzisa isilawuli sokufinyelela esisekelwe kumaka. Ungasebenzisa omaka ukuze ulawule ukufinyelela ngaphakathi kwama-akhawunti akho e-AWS. Ku-Amazon Transcribe, omaka bangangezwa emisebenzini yokubhala, amagama angokwezifiso, izihlungi zamagama angokwezifiso, namamodeli olimi angokwezifiso.

Ukuzijwayeza Okungcono Kakhulu 6 - Sebenzisa amathuluzi wokuqapha we-AWS. Ukuqapha kuyingxenye ebalulekile yokugcina ukwethembeka, ukuphepha, ukutholakala, nokusebenza kwe-Amazon Transcribe kanye nezixazululo zakho ze-AWS. Ungakwazi qapha i-Amazon Transcribe usebenzisa i-AWS CloudTrail futhi I-Amazon CloudWatch.

Ukuzijwayeza Okungcono Kakhulu 7 - Nika amandla I-AWS Config. I-AWS Config ikuvumela ukuthi uhlole, uhlole, futhi uhlole ukulungiselelwa kwezisetshenziswa zakho ze-AWS. Usebenzisa i-AWS Config, ungabuyekeza izinguquko ekucushweni nasebudlelwaneni phakathi kwezinsiza ze-AWS, uphenye imilando enemininingwane yokucushwa kwezisetshenziswa, futhi unqume ukuthobela kwakho konke ngokuqhathanisa nezilungiselelo ezicaciswe kuzinkombandlela zakho zangaphakathi. Lokhu kungakusiza wenze kube lula ukuhlola ukuthobela imithetho, ukuhlaziya ukuphepha, ukuphathwa koshintsho, nokuxazulula izinkinga zokusebenza.

Ukuqinisekisa ukuthobelana kwe-Amazon Transcribe

Izinhlelo zokusebenza ozakha ku-AWS zingase zibe ngaphansi kwezinhlelo zokuthobelana, ezifana ne-SOC, PCI, FedRAMP, ne-HIPAA. I-AWS isebenzisa abacwaningi mabhuku bezinkampani zangaphandle ukuze bahlole izinsizakalo zayo ukuze zihambisane nezinhlelo ezihlukahlukene. I-Artifact ye-AWS ikuvumela ukuba landa imibiko yokuhlolwa kwezinkampani zangaphandle.

Ukuze uthole ukuthi ingabe isevisi ye-AWS ingaphakathi kohlelo lwezinhlelo ezithile zokuthobela, bheka ku Amasevisi e-AWS ku-Scope by Compliance Programme. Ukuze uthole ulwazi olwengeziwe nezinsiza ezihlinzekwa yi-AWS ukusiza amakhasimende ngokuthobela, bheka Ukuqinisekisa ukuthobelana kwe-Amazon Transcribe futhi Izinsiza zokuthobela i-AWS.

Isiphetho

Kulokhu okuthunyelwe, ufunde ngezindlela ezihlukahlukene zokuphepha, izinqubo ezihamba phambili, namaphethini ezakhiwo atholakalayo ukuze wakhe izinhlelo zokusebenza ezivikelekile nge-Amazon Transcribe. Ungakwazi ukuvikela idatha yakho ebucayi kokubili kwezokuthutha nalapho uphumule ngokubethela okuqinile. Ukuhlelwa kabusha kwe-PII kungasetshenziswa ukunika amandla ukususwa kolwazi lomuntu siqu emibhalweni yakho uma ungafuni ukulucubungula nokuyigcina. Amaphoyinti okugcina e-VPC kanye ne-Direct Connect akuvumela ukuthi usungule ukuxhumana okuyimfihlo phakathi kwesicelo sakho nesevisi ye-Amazon Transcribe. Siphinde sahlinzeka ngezithenjwa ezizokusiza ukuthi uqinisekise ukuthobela isicelo sakho usebenzisa i-Amazon Transcribe ngezinhlelo ezifana ne-SOC, PCI, FedRAMP, ne-HIPAA.

Njengezinyathelo ezilandelayo, hlola Ukuqalisa nge-Amazon Transcribe ukuqala ngokushesha ukusebenzisa isevisi. Bukela ku Imibhalo ye-Amazon Transcribe ukuze ungene ujule emininingwaneni yesevisi. Futhi ulandele I-Amazon Transcribe ku-AWS Machine Learning Blog ukuze uhlale unolwazi lwakamuva ngamakhono amasha nokusebenzisa amacala e-Amazon Transcribe.

---

Mayelana Nombhali

Alex Bulatkin unguMklami Wezixazululo kwa-AWS. Uyakujabulela ukusiza abahlinzeki bezinsizakalo zokuxhumana bakhe izixazululo ezintsha ku-AWS ezichaza kabusha imboni yezokuxhumana. Unentshisekelo yokusebenza namakhasimende ekuletheni amandla ezinsiza ze-AWS AI ezinhlelweni zawo. U-Alex uzinze endaweni yedolobha lase-Denver futhi uthanda ukugibela intaba, ukushushuluza eqhweni, kanye ne-snowboard.

• I-SEO Powered Content & PR Distribution. Khuliswa Namuhla.
• I-PlatoData.Network Vertical Generative Ai. Zinike Amandla. Finyelela Lapha.
• I-PlatoAiStream. I-Web3 Intelligence. Ulwazi Lukhulisiwe. Finyelela Lapha.
• I-PlatoESG. Ikhabhoni, I-CleanTech, Amandla, Environment, Ilanga, Ukuphathwa Kwemfucuza. Finyelela Lapha.
• I-PlatoHealth. I-Biotech kanye ne-Clinical Trials Intelligence. Finyelela Lapha.
• Source: https://aws.amazon.com/blogs/machine-learning/best-practices-for-building-secure-applications-with-amazon-transcribe/