Umwandiko usanzwe wo kuvuga
Guhindura inyandiko ni intambwe yimbere-yandika inyandiko mbisi yanditse mumagambo yavuzwe neza mbere yuko sisitemu yo kuvuga ibivuga.
Incamake
It is what turns '$5' into 'five dollars' and '12/5/2024' into a spoken date, and getting it wrong is one of the most jarring TTS failures.
Kwibira cyane
Inyandiko yanditse yuzuyemo amagambo adasanzwe: imibare, ifaranga, amatariki, ibihe, amagambo ahinnye, URL, nibimenyetso ntawe ubivuga uko byakabaye. Ubusanzwe inyandiko (rimwe na rimwe yitwa TN imbere-iherezo) yagura ibi muburyo bwabo bwo kuvuga kugirango moderi yo hasi imenye icyo ivuga - '$ 5' ihinduka 'amadorari atanu,' 'Dr.' ihinduka 'umuganga' cyangwa 'gutwara' bitewe n'imiterere, kandi 'IV' ishobora kuba 'ine,' 'imitsi,' cyangwa inyuguti 'I-V.' Sisitemu gakondo ikoresha amategeko yandikishijwe intoki hamwe nuburemere bwa leta ya transducers (WFSTs), yizewe kandi igenzurwa. Uburyo bushya bukoresha imiterere yuburyo bukurikirana, ariko TN itunganijwe neza irashobora gutanga amakosa ateye akaga (kuvuga umubare utari wo), sisitemu yo kubyaza umusaruro rero ikoresha ibishushanyo mbonera bifite amategeko nkizamu. Context-sensitivite ni igice gikomeye: ikimenyetso kimwe kivuga mu buryo butandukanye bitewe nibidukikije.
Ubushishozi
Ubusanzwe classique isanzwe ibanza kwerekana kandi ikanashyira buri kimenyetso mubyiciro bya semotic (kardinali, icumi, itariki, amafaranga, igipimo, amagambo ahinnye), hanyuma igakoresha icyiciro cyihariye cyo kuvuga, akenshi cyubatswe nka transducer iremereye ya leta yihuta kandi igenzurwa byuzuye. Ibimenyetso bidasobanutse bitandukanijwe ukoresheje imiterere yaho hamwe nigice-cy-imvugo. Sisitemu ya Neural na Hybrid ikora nk'inyandiko-ku-kwandika ariko ikabuza ibisubizo - urugero, gutwikira ikibonezamvugo cyangwa 'gutondeka noneho kwaguka' - kugirango wirinde amakosa atemewe nko gusoma umwaka nka numero ya terefone.
Ingaruka z'Ingamba
Kugera no kugera
Itezimbere kugerwaho binyuze mu kwandukura, kuvuga, no guhuza amajwi.
Igiciro na bije
Amatsinda yibitangazamakuru arashobora kohereza amajwi yihuse hamwe na bije nto.
Umuvuduko n'igipimo
Sisitemu ireba abakiriya irashobora gutunganya imikoranire ivugwa murwego runini.
Ejo hazaza h'umwandiko usanzwe wo kuvuga
Ubusanzwe bugenda bugana kuri neural-na-amategeko ya Hybride ikomeza umutekano wikibonezamvugo cya leta itagira ingano mugihe ukoresheje urugero rwize kugirango ukemure imiterere, hiyongereyeho imvugo nini yerekana imvugo itesha umutwe, inyandiko-nyayo nindimi nyinshi icyarimwe. Ubushakashatsi bwibanze ku gukuraho amakosa 'adasubirwaho' no kuri TN indimi nyinshi aho umubare, itariki, n'amasezerano y'ifaranga bitandukanye cyane. Nkuko iherezo-iherezo rya TTS ryinjiza imirimo myinshi yimbere-iherezo, iteganya ko ibisanzwe bikomeza kugenzurwa, kugenzurwa neza kuko amakosa hano aragaragara kandi ahenze.
Gushyira mu bikorwa Isi
Gusoma '$ 1,250.50' mu ijwi riranguruye ngo 'igihumbi magana abiri na mirongo itanu n'amadorari mirongo itanu' mu mufasha w'ijwi rya banki.
Kwagura amagambo ahinnye rero 'Mutagatifu' ivugwa nk 'umuhanda' cyangwa 'umutagatifu' bitewe n'imiterere yo kugendagenda.
Kuvuga amatariki, ibihe, numero za terefone neza muri kalendari no kwibutsa porogaramu.
Guhindura ibimenyetso nibice nka '5 km' cyangwa '%' mumagambo yavuzwe kubasomyi ba ecran nibikoresho bigerwaho.
Ingaruka & Kurinda
Gukoresha nabi amajwi no kwigira ibyago byiyongera mugihe uruhushya rubuze.
Ukuri kurashobora kugabanuka hejuru yimvugo, imvugo, cyangwa urusaku rwibidukikije.
Amajwi yubukorikori arashobora kwibeshya kumvugo yukuri nta kirango gisobanutse.
Igishushanyo mbonera
Shaka uruhushya rusobanutse rwo gufata amajwi, gukoroniza, no gukoresha.
Ikizamini cyiza mubiganiro bitandukanye hamwe nuburyo bwimbere.
Sobanura igihe umuntu agomba gusuzuma cyangwa kwemeza ibisubizo.
Andika amajwi yubukorikori kandi ugumane inyandiko zerekana kubazwa.
Komeza Ubushakashatsi
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Text Normalization for Speech quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Ubuyobozi bukurikira
Umwandiko Kuvuga
Ibibazo bikunze kubazwa
What is Text Normalization for Speech?
Guhindura inyandiko ni intambwe yimbere-yandika inyandiko mbisi yanditse mumagambo yavuzwe neza mbere yuko sisitemu yo kuvuga ibivuga. Nicyo gihindura '$ 5' mo 'amadorari atanu' na '12 / 5/2024 'mu itariki yavuzwe, kandi kubyibeshya ni kimwe mu byananiye TTS.
Ni ubuhe buryo busanzwe bwo guhindura imvugo bukora?
Ubusanzwe bwagura ibimenyetso bitari bisanzwe nkimibare, amatariki, hamwe nincamake muburyo bwabo bwo kuvuga mbere yo guhuza.
Nigute sisitemu nziza igomba guhuza '$ 5' kugirango tuvuge?
Ifaranga risaba gutondekanya no kuvuga mu magambo ikimenyetso, bityo '$ 5' ivugwa nk '' amadolari atanu, 'ntabwo ibumoso cyangwa iburyo.
Ni ukubera iki guhuza ikimenyetso nka 'Dr.' cyangwa 'IV' amayeri?
'Dr.' irashobora kuba 'umuganga' cyangwa 'gutwara' na 'IV' irashobora kuba 'ine,' 'imitsi,' cyangwa inyuguti - imiterere igena imvugo iboneye.
Ni ubuhe buhanga gakondo bukoreshwa cyane mu kubaka amategeko yizewe, yumvikana?
WFSTs ikubiyemo ikibonezamvugo cyakozwe n'intoki cyihuta kandi kigenzurwa neza, bigatuma bahitamo igihe kirekire kubyara TN.
Ni ukubera iki sisitemu yo kubyaza umusaruro akenshi irinda imyandiko isanzwe itanduye?
Imyitwarire idahwitse ya TN irashobora gutanga umusaruro wizewe ariko uteje akaga (urugero, ishusho itari yo), amategeko rero akora nk'izamu muri sisitemu ya Hybrid.