InoteveraGaidhi rinotevera
MelGAN Generative Vocoder
AI Yenyaya
Chidzidzo cheAI yeRuzha
Parallel WaveGAN Vocoder
Parallel WaveGAN inokurumidza neural vocoder inoshandura mel-spectrogram kuita yakaomeswa waveform uchishandisa diki GAN, ichigadzira ese masampuli kamwechete.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
Izvo zvine basa nekuti inopa pedyo-chaiyo-nguva, yemhando yepamusoro kutaura ine compact modhi.
Kudzika Kwakadzika
Vocoder ndiyo nhanho yekupedzisira yepombi yeTTS: inoshandura acoustic chimiro mepu (kazhinji mel-spectrogram) kuita iyo chaiyo ruzha wave iwe yaunonzwa. Parallel WaveGAN, yakakurudzirwa naYamamoto, Rwiyo, naKim muna 2019, inoita izvi neisina-autoregressive WaveNet-maitiro jenareta akadzidziswa seanogadzira adversarial network. Panzvimbo pekufanotaura sampuli imwe yekuteerera panguva seyekutanga WaveNet, inoburitsa iyo waveform yese yakafanana, ichiita kuti ikurumidze kukurumidza. Yayo kiyi yekubikira inosanganisa kurasikirwa kweanopikisa pamwe neakawanda-resolution pfupi-nguva pfupi-pfupi Fourier shanduko (STFT) kurasikirwa, saka modhi inoenderana neiyo chaiyo chiratidzo kuyambuka akati wandei nguva uye frequency zviyero. Mhedzisiro idiki jenareta (yakatenderedza miriyoni 1.4 paramita) inomhanya kakawanda nekukurumidza kupfuura nguva chaiyo paGPU.
Kuitwa muNyika yeChokwadi
Chaiyo-nguva yekutaura inobuda mune nharembozha vabatsiri uko latency uye modhi saizi zvine basa
Kushanda sewaveform jenareta rakapetwa neacoustic modhi seTacotron 2 kana FastSpeech
Pa-mudziyo mameseji-ku-kutaura ezvishandiso zvekusvikika zvisingagone kuvimba negore
Mazwi ekushandura masisitimu anogadzirazve akashandurwa ma spectrogram kuita echisikigo-inonzwika odhiyo
Ramangwana reParallel WaveGAN Vocoder
Parallel WaveGAN yakabatsira kumisikidza GAN vocoders seyazvinoita default, uye yayo yakawanda-resolution STFT kurasikirwa ikozvino kwave kuoneka kune vese vanotsiva seHiFi-GAN uye akawanda ekutepfenyura masisitimu. Iyo trajectory inonongedza kune anogara ari madiki, akaderera-latency mavokoder e-pa-mudziyo vabatsiri, zvinoyamutsira kunzwa, uye mhenyu shanduko yezwi, pamwe nevokodha yepasirese inoenda kune vasingaonekwe vatauri. Tarisira kubatanidzwa kwakasimba nekuguma-kusvika-kumagumo TTS uye kutumirwa kunoshanda pane nharembozha uye yakamisikidzwa machipisi.
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 Parallel WaveGAN Vocoder 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 Parallel WaveGAN Vocoder?
Parallel WaveGAN inokurumidza neural vocoder inoshandura mel-spectrogram kuita yakaomeswa waveform uchishandisa diki GAN, ichigadzira ese masampuli kamwechete. Izvo zvine basa nekuti inopa pedyo-chaiyo-nguva, yemhando yepamusoro kutaura ine compact modhi.
Nderipi basa revokoda seParallel WaveGAN?
Vocoder ndiyo yekupedzisira TTS nhanho inogadzira iyo chaiyo ruzha wave kubva kune acoustic maficha senge mel-spectrogram.
Ko Parallel WaveGAN inogadzira sei maodhiyo sampuli ichienzaniswa neyekutanga WaveNet?
Parallel WaveGAN haina-autoregressive, inogadzira iyo waveform yese kamwechete pane sampuli nemuenzaniso, izvo zvinoita kuti ikurumidze.
Ndeipi kurasikirwa kuri pakati peParallel WaveGAN kunze kwekurasikirwa kweanopikisa?
Inobatanidza kurasikirwa kwemhandu pamwe ne-multi-resolution STFT kurasikirwa kunoenzanisa spectrogram magnitudes pazvikero zvakawanda.
Ndeipi dhizaini inoshandiswa neParallel WaveGAN jenareta?
Iyo jenareta ndeye WaveNet-senge network yakavakirwa kubva kune yakadzikama convolutions, yakarongedzwa pane mel-spectrogram uye ruzha.
Nei Parallel WaveGAN ichikwezva kuendeswa?
Iine angangoita miriyoni miriyoni paramita idiki uye inomhanya kakawanda nekukurumidza kupfuura nguva chaiyo, yakanakira kushandiswa pa-mudziyo.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi