AI Lip Sync Video Generator

Soo dejisan sawirka wajiga iyo audio clip — hesho video hadal-hore la dhab ah buskudka sync, madaxa pose, iyo indho-indheynta. Powered by SadTalker (MIT). Commercial isticmaalka OK.

Ma lihin codadka TTS ee afkaaga weli. Na caawi inaad ku darto kuwaaga! Iibso Codkaaga

Soo qaado Face + Audio

1,000 xarfo daqiiqad kasta

Riix & riix faylka halkan, ama booqo

JPG, PNG, or short MP4/WebM. Max 10MB. One clear, well-lit face works best.

Faylka.mp3

0 MB

Riix & riix faylka halkan, ama booqo

MP3, WAV, M4A, or FLAC. Max 10MB. Free: up to 30 sec. Pro: up to 5 min.

Faylka.mp3

0 MB

La socodsiinayo...

Rendering video. Tani caadi ahaan qaadataa 30 ilbiriqsi ilaa 2 daqiiqo.

Fiidiyowgaaga hadalka-hore

Soo deji

SadTalker

SadTalker (CVPR 2023, Tencent ARC) waa mid furan-source hadalka-hore ee qaabka oo animates sawir weji kaliya in ay ka hadlaan wax kasta oo audio. Wav2Lip kala duwan ka duwan, SadTalker sidoo kale animates madaxa pose, blinks, iyo muujinta natiijo ka badan oo dabiici ah.

Koodka iyo miisaanka waa MIT-liisan dhamaadka ilaa dhamaadka - ma aha Llama, Gemma, ama aan ganacsi ahayn ee madaxa - sidaas darteed videos aad abuurto waa ammaan u ah isticmaalka ganacsi.

Talooyin si ay u hesho natiijooyinka ugu fiican

  • isticmaal sawir tayo sare leh oo si fiican loo iftiimiyay - indhaha ayaa muuqata, afka waa la xiray
  • Wajiga dhexe, square ama 4: 5 qaabka ratio shaqooyinka ugu fiican
  • Dhageysiga hadalka ee nadiif ah (ma jiro muusig) wuxuu keenaa isku dheelitirnaan buskud oo adag
  • Gaar u ah GFPGAN for hero shots — laba jeer u soo bandhigaan waqti laakiin sharraxaadda sharraxaadda
  • isticmaal Still preset markaad rabto avatar xasilloon oo la qaaday

Libaax Sync Video qorshayaasha

Bilaash u bilow, kor u qaad markaad u baahan tahay in ka badan

Bilaash
  • 30-sekondii xaddidaadda audio
  • 256 px natiijooyinka
  • "Still" oo kaliya
  • Wax soo saar
Ugu caansan
Xisaab Bilaash ah
  • 30-sekondii xaddidaadda audio
  • Labada "full" iyo "ha joogin" oo hore loo dhigay
  • 256 / 512 pixels soosaarka
  • GFPGAN wajiga kordhinta
Ka diiwaangashan Free
Pro
  • 5-daqiiqo xaddi audio ah
  • Xididka GPU ee Horumarka
  • API access (multipart soo dejinta)
  • Webhook dhamaystirka callbacks
  • isticmaal ganacsi (lacag la'aan MIT)
Kordhi

Su'aalaha badanaa la waydiiyo

Soo dejisan sawir waji iyo clip audio, iyo AI soo saartaa video ah ee wejigaas ku hadlaya audio la dhaqdhaqaaqyada buskudka dhabta ah, madaxa, iyo indho-indheynta. Built on SadTalker (CVPR 2023), a MIT-liisan hadalka-hore ee qaabka oo muujinaya muuqaalka in ka badan qaabka afka.

Wajiga soo dejinta waxaa laga yaabaa in JPG ama PNG sawir (ugu badnaan 10 MB) ama MP4 / WebM video socday gaaban (waxaan isticmaalnaa frame ugu horeysay). Dhagaxa socday waxaa laga yaabaa in MP3, WAV, M4A, ama FLAC ilaa 10 MB. Waxaan resamply audio in 16 kHz gudaha.

xisaabaadka bilaashka ah: ilaa 30 ilbiriqsi oo clip ah. isticmaalayaasha bixiyo: ilaa 5 daqiiqo oo codsi ah. audio dheer ka dhigaysa waqti sii dheer oo ka dhigaysa iyo qiimaha persona sare.

Lip sync video isticmaalaa 1,000 xarafka daqiiqadii video soo saaro. A 30-second clip = 30,000 xarafka. Qiimaha waa la billaabay hore ka mid ah miisaanka aad character iyo dib loo soo celiyo si otomaatig ah haddii soosaarka fashilmo.

Haa — SadTalker code iyo miisaanka waa MIT licensed dhamaadka ilaa dhamaadka (no Llama, Gemma, ama aan ganacsi ah). videos aad abuurto waa adiga inaad isticmaali ganacsiga. Waxaad mas'uul ka tahay in ay leeyihiin xuquuqda sawirka wajiga asalka ah iyo audio aad soo dejisan.

Ku saabsan 30 ilbiriqsi oo 5-second clip ah server-keena A100, oo si toos ah u sii kordhaya oo leh dhererka audio. Markaad awood u yeelatid GFPGAN wajiga, waxay si toos ah u kordhisaa waqtiga soo bandhigida laakiin waxay soo saartaa wax soo saar tayo sare leh.

Full preset (default) animates madaxa pose, blinks, iyo muujinta la buskudka, soo saarka video badan oo dabiici ah hadalka-hore. weli preset xiran madaxa meel iyo animates kaliya afka — faa'iido leh marka aad rabto in aad avatar isku dheeli tiran toogtay.

GFPGAN waa qaabka soo celinta wajiga oo ka dhigaysa faahfaahinta wajiga ka dib markii la soo bandhigay. Waxay nadiifisaa waxyaabo iyo 256-pixel soo saarka u egtahay 512. Waxay si toos ah u kordhisaa waqtiga soo bandhigida laakiin waa ku habboon tahay sawirka aabayaasha.

SadTalker soo bandhigaa at 256 px by default. Isku beddel 512 px cabbirka u sharraxan soo saarka (badnaan, VRAM sare) ama awood u GFPGAN enhancer in upscale faahfaahinta wajiga. Si aad u hesho natiijooyinka ugu fiican, soo dejisan tayo sare leh, sawirka sawirka wanaagsan.

Haa. soo dejisan MP4 ama WebM sida wejiga soo dejinta oo waxaan isticmaali doonaa frame-ka hore sida aqoonsiga wadista. Si aad u buuxdo video dib-dubbing (per-frame afka bedelka), eeg soo socda Dubbing Studio video pipeline.

Haa. POST codsiga multipart in /api/v1/lipsync/ la wajiga iyo audio goobaha, ka dibna codbixinta /api/v1/lipsync/result/?uuid = ilaa xaaladda waa "dhamaaday". jawaabta waxaa ku jira URL in MP4 soo bandhigay. API helitaanka u baahan yahay qorshe la bixiyo.

SadTalker isticmaalaa wajiga-u-qaabeynta si ay u ogaato oo u beerto weji ugu caansan. Natiijooyinka ugu fiican, soo dejisan sawir la mid ah qof mid ka mid ah, indhaha muuqata, iyo occlusion ugu yar. Group photos waxay keeni kartaa natiijooyinka aan la saadaalin karin.
5.0/5 (1)

Maxaa aan ku hagaajin karnaa? Jawaabtaada waxay naga caawisaa inaan xallino dhibaatooyinka.

Miyaad diyaar u tahay inaad bilowdo?

Ka diiwaangashan oo bilaash ah oo ka heli 15,000 xaraf. No credit card loo baahan yahay.