NVIDIA Riva da NeMo Magana
NVIDIA Riva shine SDK mai haɓaka GPU don samar da magana AI (ASR, TTS, da fassarar), yayin da NeMo shine kayan aikin buɗe tushen don horarwa da daidaita samfuran da ke ƙasa.
Dubawa
Together they let developers build fast, customizable voice applications that run on NVIDIA hardware.
Zurfafa nutsewa
NeMo (Modules Neural) shine tsarin buɗaɗɗen tushen tushen PyTorch na NVIDIA don gina AI na tattaunawa. Yana jigilar samfuran da aka riga aka horar don gane magana ta atomatik (ASR), rubutu-zuwa-magana (TTS), da ayyukan harshe na halitta, waɗanda aka tsara azaman 'na'urorin jijiyoyi' waɗanda za'a iya sake amfani da su za ku iya daidaitawa akan bayanan ku. Riva shine gefen turawa: yana tattara ingantattun samfura a bayan sabar gRPC mai yawo, ta amfani da TensorRT da Triton Inference Server don buga ƙarancin latency a sikelin. Aiki na yau da kullun yana ba da horo ko daidaita samfuri a cikin NeMo, yana fitar da shi zuwa tsarin Riva, sannan ya yi amfani da shi don rubutawa ko haɗawa ta ainihi. Riva yana goyan bayan ƙaddamar da yawo tare da tambura matakin kalma, muryoyin TTS na jijiyoyi, diarization na lasifikar, da kuma yaruka da yawa, duk an daidaita su don gudanar da ingantaccen aiki akan NVIDIA GPUs.
Fahimtar Fasaha
Gudun Riva ya fito ne daga haɗa samfura tare da TensorRT da yi musu hidima ta hanyar Triton, wanda ke haɗa kernels, ya shafi gauraye-daidai (FP16/INT8), da buƙatun buƙatun lokaci guda. Samfuran ASR kamar Conformer-CTC ko Parakeet rafi mai jiwuwa cikin ƙananan guntu yayin kiyaye mahallin, samar da juzu'i a cikin dubun milliseconds. Bututun TTS sun haɗa samfurin sauti (misali, FastPitch) tare da vocoder na jijiyoyi (misali, HiFi-GAN) don samar da sifofin igiyoyi cikin sauri fiye da ainihin lokacin akan GPU guda.
Dabarun Tasiri
Shiga ku isa
Yana inganta samun dama ta hanyar rubutu, ba da labari, da mu'amalar murya.
Kudin da kasafin kuɗi
Ƙungiyoyin kafofin watsa labaru na iya jigilar sauti mai gogewa cikin sauri tare da ƙaramin kasafin kuɗi.
Gudu da sikelin
Tsarin fuskantar abokin ciniki na iya aiwatar da hulɗar magana a mafi girman ma'auni.
Makomar NVIDIA Riva da Magana NeMo
NVIDIA tana tura Riva da NeMo zuwa ga mafi girma, ƙarin ƙirar magana mai tushe da yawa da haɗin kai tare da wakilai na tushen LLM don mataimakan murya na ƙarshe zuwa ƙarshe. Yi tsammanin ingantaccen gyare-gyare (ƙarfafa kalmomi, muryoyin al'ada daga mintuna na bayanai), ingantaccen yanayi mai ƙarfi, da turawa wanda ya mamaye GPUs na cibiyar bayanai zuwa na'urori kamar Jetson. Kamar yadda NeMo ke tasowa tare da ƙirar ƙira, layi tsakanin fahimtar magana, fassarar, da tunanin tattaunawa zai ci gaba da yin duhu cikin bututun da aka haɗa na ainihin lokaci.
Aiwatar da Gaskiyar Duniya
Rubutun wurin kira na ainihi da wakili mai rai yana taimakawa wannan taken kiran abokin ciniki tare da tambura matakin-kalmar
Gina masu alamar muryoyin TTS na al'ada don mataimaki na zahiri ta hanyar daidaita FastPitch a cikin NeMo akan ƴan sa'o'i na rikodi.
Taken kai tsaye da fassarar magana don taron taron bidiyo ko yawo a kan NVIDIA GPUs
Kyakkyawan daidaita samfurin ASR na Conformer akan ƙayyadaddun likitancin yanki ko ƙamus na doka ta amfani da NeMo, sannan yi masa hidima ta Riva
Hatsari & Tsare-tsare
Rashin amfani da murya da haɗarin kwaikwaya yana ƙaruwa lokacin da aka rasa izini.
Daidaituwa na iya faɗuwa cikin lafuzza, yaruka, ko mahalli masu hayaniya.
Ana iya kuskuren sauti na roba don ingantacciyar magana ba tare da bayyananniyar lakabi ba.
Taswirar Hanya
Sami tabbataccen izini don ɗaukar murya, cloning, da sake amfani.
Gwajin ingantattun masu magana daban-daban da yanayin baya.
Ƙayyade lokacin da dole ne ɗan adam ya duba ko ya amince da abubuwan da aka fitar.
Yi lakabin sauti na roba da kuma adana bayanan da aka tabbatar don yin lissafi.
Ci gaba da Bincike
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 NVIDIA Riva and NeMo 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
Jagora na gaba
Ma'aunin ingancin Magana na PESQ da STOI
Tambayoyin da ake yawan yi
What is NVIDIA Riva and NeMo Speech?
NVIDIA Riva shine SDK mai haɓaka GPU don samar da magana AI (ASR, TTS, da fassarar), yayin da NeMo shine kayan aikin buɗe tushen don horarwa da daidaita samfuran da ke ƙasa. Tare suna ƙyale masu haɓakawa su yi sauri, aikace-aikacen murya da za a iya daidaita su waɗanda ke gudana akan kayan aikin NVIDIA.
Menene babban bambanci tsakanin NeMo da Riva?
NeMo shine kayan aikin buɗaɗɗen tushen kayan aiki don ginawa da daidaita magana da ƙirar harshe, yayin da Riva fakitin kuma yana hidima ga waɗannan samfuran don amfanin samar da ƙarancin jinkiri.
Wadanne fasahohin NVIDIA guda biyu ne Riva ke dogaro da su don cimma ƙarancin latency?
Riva yana tattara samfura tare da TensorRT don ingantaccen aiwatar da GPU kuma yana yi musu hidima ta Triton Inference Server, wanda ke sarrafa batching mai ƙarfi da daidaituwa.
A cikin bututun Riva TTS na yau da kullun, menene aikin vocoder kamar HiFi-GAN?
Samfurin acoustic kamar FastPitch yana annabta fasalulluka daga rubutu, kuma muryar murya kamar HiFi-GAN tana juya waɗannan fasalulluka zuwa yanayin ƙarar sauti na ƙarshe.
Menene 'streaming' ASR a cikin Riva ke kunna?
Ganewar yawo yana sarrafa sauti cikin ƙananan guntu kuma yana fitar da sakamako na ɗan lokaci a kusa da ainihin lokaci, maimakon jiran gabaɗayan furucin ya ƙare.
Wanne misali ne na keɓancewa NeMo ke ba da damar ƙirar magana?
NeMo yana ƙyale masu haɓakawa su daidaita ƙirar da aka riga aka horar da su akan bayanan nasu, misali daidaita ƙirar ASR don gane ƙwararrun likita ko ƙamus na doka.