Nhazi ederede maka okwu
Nhazi ederede bụ nzọụkwụ dị n'ihu nke na-edegharị ederede ederede ka ọ bụrụ okwu ekwupụtara nke ọma tupu usoro okwu ekwu ya.
Nchịkọta
It is what turns '$5' into 'five dollars' and '12/5/2024' into a spoken date, and getting it wrong is one of the most jarring TTS failures.
Ime miri emi
Ederede edere juputara na okwu na-abụghị ọkọlọtọ: ọnụọgụ, ego, ụbọchị, oge, mkpirisi, URL, na akara nke ọ nweghị onye na-akpọpụta n'ụzọ nkịtị. Ederede normalization (mgbe ụfọdụ a na-akpọ TN n'ihu-ọgwụgwụ) na-agbasawanye ndị a n'ụdị ha na-ekwu okwu ka ihe atụ nke ala mara ihe ọ ga-ekwu n'ezie - '$ 5' na-aghọ 'dollar ise,'' Dr. na-aghọ 'dọkịta' ma ọ bụ 'drive' dabere na onodu ya, na 'IV' nwere ike ịbụ 'anọ,' 'intravenous,' ma ọ bụ mkpụrụedemede 'I-V'. Usoro ọdịnala na-eji iwu ejiri aka dee yana ndị ntụgharị steeti oke oke (WFSTs), bụ nke a pụrụ ịdabere na ya na nke a ga-enyocha ya. Ụzọ ọhụrụ na-eji usoro neural usoro-na-usoro ụdị, ma dị ọcha neural TN nwere ike ịmepụta ize ndụ njehie (na-ekwu na-ezighị ezi nọmba), otú mmepụta usoro na-ejikarị ngwakọ aghụghọ na iwu dị ka guardrails. Ọdịiche dị n'echiche bụ akụkụ siri ike: otu akara ahụ na-ekwupụta n'ụzọ dị iche dabere na gburugburu ya.
Nghọta nka nka
Omuma nkịtị na-ebu ụzọ mee ka tokenize ma na-ekewa akara ọ bụla ka ọ bụrụ klaasị semiotic (kadinal, decimal, ụbọchị, ego, nha, mkpirisi), wee na-etinye verbalizer nke klaasị akọwapụtara, nke a na-ewukarị dị ka onye ntụgharị steeti nwere oke arọ nke na-adị ngwa ngwa na nke a na-enyocha ya nke ọma. A na-emegharị akara ngosi enweghị mgbagha site na iji ọnọdụ mpaghara na akụkụ nke okwu. Sistemụ akwara na ngwakọ na-ahazi ya dị ka idegharị ederede na ederede mana ọ na-egbochi ya - dịka ọmụmaatụ, ikpuchi ụtọ asụsụ ma ọ bụ 'ịkpado wee gbasaa' - iji gbochie mmejọ anabataghị dị ka ịgụ otu afọ dịka nọmba ekwentị.
Mmetụta atụmatụ
Nweta na iru
Ọ na-eme ka nnweta ya dịkwuo mma site na ndegharị, ịkọ akụkọ, na ntụgharị olu.
Ọnụ ego na mmefu ego
Ndị otu mgbasa ozi nwere ike ibubata ọdịyo a na-egbu maramara ngwa ngwa site na iji obere mmefu ego.
Ọsọ na ọnụ ọgụgụ
Sistemụ na-eche ihu ndị ahịa nwere ike hazie mkparịta ụka n'ọtụtụ buru ibu.
Ọdịnihu nke nhazi ederede maka okwu
Normalization na-aga n'ihu na ngwakọ-na-achị achị nke na-edobe nchekwa nke grammar steeti nwere njedebe mgbe ị na-eji ụdị mmụta iji dozie ọnọdụ, gbakwunyere ụdị asụsụ buru ibu nke na-ejikwa ihe na-adịghị mma, ederede ụwa na ọtụtụ asụsụ ozugbo. Nchọpụta lekwasịrị anya n'iwepụ mperi 'enweghị mgbake' yana n'ọtụtụ asụsụ TN ebe ọnụọgụ, ụbọchị na mgbakọ ego dị iche iche. Dị ka njedebe-na-ọgwụgwụ TTS na-etinyekwu ọrụ n'ihu, na-atụ anya ka ịmegharị ahụ ka ọ bụrụ ihe a na-achịkwa, nke a na-enyocha kpọmkwem n'ihi na mmejọ ebe a bụ ihe a na-ahụ anya ma dị ọnụ.
Mmejuputa n'ezie n'ụwa
Ịgụ '$1,250.50' n'olu dara ụda dị ka 'otu puku dollar na narị abụọ na iri ise' na onye enyemaka olu ụlọ akụ.
Na-amụba mkpirisi mkpirisi ka 'St.' a na-asụ dị ka 'okporo ámá' ma ọ bụ 'senti' dabere na gburugburu na mkpagharị igodo.
Ịkọpụta ụbọchị, oge, na nọmba ekwentị nke ọma na kalịnda na ngwa nchetara.
Na-atụgharị akara na nkeji dị ka '5 km' ma ọ bụ '%' n'ime okwu ekwuru maka ndị na-agụ ihuenyo na ngwaọrụ nnweta.
Ihe ize ndụ & okporo ụzọ nche
Iji olu eme ihe na ihe egwu mpụta ga-abawanye mgbe nkwenye na-efu.
Izi ezi nwere ike ịdaba n'ofe ụda olu, olumba ma ọ bụ gburugburu mkpọtụ.
Enwere ike imehie ọdịyo sịntetik dị ka ezigbo okwu na-enweghị akara doro anya.
Map mmejuputa
Nweta nkwenye doro anya maka ijide olu, imechi, na ijigharị.
Nwale ogo n'ofe ndị na-ekwu okwu dị iche iche yana ọnọdụ ndabere.
Kọwaa mgbe mmadụ ga-enyocha ma ọ bụ kwado nsonye.
Deba aha ọdịyo sịntetik ma debe ndekọ ihe ndekọ maka ịza ajụjụ.
Nọgide na-eme nchọpụta
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
Ntuziaka na-esote
Ederede gaa na Okwu
Ajụjụ a na-ajụkarị
What is Text Normalization for Speech?
Nhazi ederede bụ nzọụkwụ dị n'ihu nke na-edegharị ederede ederede ka ọ bụrụ okwu ekwupụtara nke ọma tupu usoro okwu ekwu ya. Ọ bụ ihe na-atụgharị '$5' ka ọ bụrụ 'dolla ise' na '12/5/2024' ka ọ bụrụ ụbọchị a na-ekwu, na ịnweta ya na-ezighi ezi bụ otu n'ime ọdịda TTS kacha njọ.
Kedu ihe nhazi ederede maka okwu bụ isi na-eme?
Normalization na-agbasawanye akara ndị na-abụghị ọkọlọtọ dị ka ọnụọgụgụ, ụbọchị, na mkpirisi n'ime ụdị nkwuwa okwu ha tupu njikọ.
Kedu ka usoro dị mma kwesịrị isi mee ka '$5' dị mma maka okwu?
Ego chọrọ ịhazigharị na ịkọwapụta akara ngosi ahụ, ya mere a na-ekwu '$5' ka 'dolla ise,' ọ bụghị n'ụzọ nkịtị aka ekpe gaa n'aka nri.
Kedu ihe kpatara na-edozi akara dị ka 'Dr.' ma ọ bụ 'IV' dị aghụghọ?
'Dr.' nwere ike ịbụ 'dọkịta' ma ọ bụ 'drive' na 'IV' nwere ike ịbụ 'anọ,' 'intravenous,' ma ọ bụ akwụkwọ ozi - ihe gbara ya gburugburu na-ekpebi nkwupụta okwu ziri ezi.
Kedu nkà na ụzụ ọdịnala a na-eji eme ihe iji wuo iwu nkịtị, nke a pụrụ ịdabere na ya?
WFST na-edekọ ụtọ asụsụ ejiri aka mee ngwa ngwa na nke a na-enyocha ya nke ọma, na-eme ka ha bụrụ nhọrọ dị ogologo maka mmepụta TN.
Kedu ihe kpatara usoro mmepụta na-ejikarị ezere nhazi ederede akwara dị ọcha?
TN neural na-enweghị mgbochi nwere ike iwepụta mpụta ntụkwasị obi mana dị egwu (dịka ọmụmaatụ, ọnụ ọgụgụ na-ezighi ezi), yabụ iwu na-arụ ọrụ dị ka ihe nchekwa na sistemu ngwakọ.