InoteveraGaidhi rinotevera
Kamba TTS Autoregressive Synthesis
AI Yenyaya
Chidzidzo cheAI yeRuzha
FastSpeech uye Non-Autoregressive TTS
FastSpeech inogadzira iyo yese yekutaura spectrogram yakafanana kwete furemu imwe panguva, ichiita synthesis inokurumidza uye yakagadzikana zvakanyanya.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
Yakagadzirisa chizvarwa chinononoka, chakakanganisika chakatambudza ekare autoregressive modhi seTacotron.
Kudzika Kwakadzika
Kare neural TTS modhi seTacotron 2 ndeye autoregressive: vanofanotaura yega yega odhiyo furemu yakamisikidzwa pane yakapfuura, inononoka uye inokaruka kusvetuka kana kudzokororwa mazwi kana kutarisisa kwatadza. FastSpeech, yakaunzwa na Microsoft neZhejiang University muna 2019, inoshandura izvi nekufanotaura mafaremu ese kamwechete. A Transformer-based feed-forward network inotora foni, inofanotaura zvakajeka kuti foni imwe neimwe inofanira kugara kwenguva yakareba sei ine regulator yehurefu, uye inowedzera kutevedzana kune nhamba chaiyo yemafuremu isati yagadzira spectrogram mune imwechete pass. FastSpeech 2 yakagadziridzwa pane izvi nekufanotaura kukwirira nesimba zvakare, uye nekudzidziswa nguva yezvinangwa kubva pakumanikidzwa kurongeka pachinzvimbo chekudzibvisa kubva kune inononoka mudzidzisi modhi, ichipa yakajairika uye inodzoreka kutaura.
Kuitwa muNyika yeChokwadi
Real-time navigation apps inogadzira inoshandura-ye-ye-kutenderedza inzwi nekukurumidza uchishandisa yakafanana FastSpeech-style synthesis.
Mutengi-sevhisi IVR masisitimu anoshandura mavara ane simba kuita kutaura pachiyero pasina zvikanganiso zvekusvetuka-mazwi.
Kuwanika kwevaverengi vezvidzitiro vanogadzira kutaura nekukurumidza, kwakavimbika kwemagwaro marefu pane zvine mwero Hardware.
Maturusi ezvemukati mezwi ngaaite kuti vagadziri vawedzere kukwirira uye chiyero chekutaura zvakananga, nekuda kweFastSpeech 2's yakajeka nzwi uye simba rekufanotaura.
Ramangwana reFastSpeech uye Non-Autoregressive TTS
Non-autoregressive synthesis ikozvino ndiyo yakasarudzika yekugadzira TTS nekuti inokurumidza, yakasimba, uye inodzoreka. Masisitimu emangwana anosundidzira akananga kune yakanakisa prosody control, yakaderera-latency kutenderera kwehupenyu maapplication, uye ekupedzisira-kusvika-kumagumo akasiyana anosvetuka yepakati spectrogram zvachose. Diffusion- uye kuyerera-kwakavakirwa kusiri-autoregressive modhi zvakare kuri kusimuka, kusanganisa FastSpeech's parallelism ine yakasimba generative mhando, nepo yakajeka pitch uye nguva yekudzora inoramba ichikosheswa kune inogadziriswa, inotaura zvigadzirwa zvezwi.
Kuzvishandisa
Nei zvichikosha
- Svika uye svika. Inonatsiridza kusvikika kuburikidza nekunyora, kurondedzera, uye mazwi ekubatanidza.
- Mutengo uye bhajeti. Zvikwata zveMedia zvinogona kutumira odhiyo yakakwenenzverwa nekukurumidza nemabhajeti madiki.
- Kumhanya uye chiyero. Masisitimu anotarisana nevatengi anogona kugadzirisa kutaurirana kwekutaura pamwero mukuru.
Chenjera
- Kushandisa izwi zvisizvo uye njodzi dzekuedzesera dzinowedzera kana chibvumirano chisipo.
- Kururama kunogona kudonha mumitauro, mataurirwo, kana nharaunda dzine ruzha.
- Synthetic audio inogona kukanganisa kutaura kwechokwadi isina mavara akajeka.
Matanho anotevera
- Wana mvumo yakajeka yekutora inzwi, kugadzira, uye kushandisa zvakare.
- Yedza mhando pavatauri vakasiyana uye mamiriro ekumashure.
- Tsanangura apo munhu anofanira kuongorora kana kubvumidza zvabuda.
- Label synthetic odhiyo uye chengetedza marekodhi ekuzvidavirira.
Ramba Uchiongorora
Tsamba yemahara
Tora iyo yemazuva ese AI pfupiso
Nhatu dzakasimbiswa dzeAI nyaya mangwanani ega ega pakati pevhiki, dzakanyorwa neChirungu chakajeka. Yemahara zvachose, hapana ads.
Imwe email zuva rega rega revhiki. Bvisa kunyoresa kamwe chete. Hatimbotengese kana kugovera kero yako.
Zviedze iwe pachako
Take the FastSpeech and Non-Autoregressive TTS quiz
Mhinduro yepakarepo pamhinduro yega yega, uye chitupa chinogovaniswa chine ID inobatika kana wapedza kosi.
Tsigira dzidzo yemahara yeAI. AI Understanding ndeye 501(c)(3) isingabatsiri - hapana ads, hapana paywall, zvachose. Ita chipo
Mibvunzo inowanzo bvunzwa
Chii chinonzi FastSpeech uye Non-Autoregressive TTS?
FastSpeech inogadzira iyo yese yekutaura spectrogram yakafanana kwete furemu imwe panguva, ichiita synthesis inokurumidza uye yakagadzikana zvakanyanya. Yakagadzirisa chizvarwa chinononoka, chakakanganisika chakatambudza ekare autoregressive modhi seTacotron.
Chii chinonzi 'chisina-autoregressive' chinorevei mumamiriro eFastSpeech?
Non-autoregressive zvinoreva kuti mafuremu anogadzirwa anoenderana mune imwe chete pass, asina kuisirwa imwe-ne-imwe pamafuremu apfuura.
Nderipi dambudziko reautoregressive modhi seTacotron 2 inoita FastSpeech inogadzirisa?
Autoregressive kutarisisa kunogona kurongeka, zvichikonzera kusvetuka kana kudzokororwa mazwi, uye sequential decoding inononoka; FastSpeech inogadzirisa zvese.
Ndeipi basa reiyo 'refu regulator' muFastSpeech?
Iyo kureba regulator inowedzera maficha efonimu neaya akafanotaura, kurongedza kupfupika kwemavara kutevedzana kune refu spectrogram.
Which additional speech variations does FastSpeech 2 condition on during training and inference?
FastSpeech 2 uses duration, pitch, and energy information as conditional inputs to better model speech variation.
Nei FastSpeech's inference nguva isingakura nekureba kwemutsara?
Nekuti chizvarwa chakafanana, kuwedzera mamwe mafuremu hakuwedzere matanho anoteedzana nenzira inoita autoregressive decoding.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi