AudioLDM 2
Free.ai (self-hosted)
·
music
·
~5000 tokens per track
·
4.0
kubva 4 1 mushandisi munyaya ino
AudioLDM 2 is a music generation model built by Haohe Liu. Strongest at Music generation from text. Self-inotarisirwa pa Free.ai GPUs - inofamba zvakasununguka pazuva rako re token pool (5,000 tokens per track). Yakaburitswa pasi pe Apache 2.0 — kushandiswa kwekutengesa kwakabvumirwa pa Free.ai.
Usati washandisa
OpenAI-inowirirana REST API. Kugadzira chiratidzo uye kukumbira iyi model mumasekondi.
curl -X POST https://api.free.ai/v1/music/generate/ \
-H "Authorization: Bearer sk-free-..." \
-H "Content-Type: application/json" \
-d '{"model":"audioldm2","prompt":"your prompt here"}'Semuenzaniso mamodheru
Zvimwe zvinobvunzwa kakawanda
AudioLDM 2 inogadzira musika wekutanga wezvipfeko (uye pazvimwe zvipenya, mavondo) kubva pabhokisi remashoko. Dzivisa pfungwa, genre, zvipfeko, kana BPM uye AudioLDM 2 inodzosera yakati wandei yezvipenya.
10 kusvika 60 masekondi pagore zvichienderana nechigadzirwa. Kuti uwane mapikicha akareba, shandisa makore akawanda ne /music/stitch/ kana shandisa 30-sekondi segments uye crossfade.
Zvimwe zvirongwa zvinogadzira mashoko kubva mumashoko aunotipa (pedza mashoko mufomu); zvimwe zvinogadzira mashoko kubva mumashoko aunotipa chete. Cheka mufomu pane iyi peji — mashoko ari mufomu zvinoreva kuti mashoko ari kutsigirwa.
Pop, rock, EDM, jazz, ambient, classical, lo-fi, hip-hop instrumentals zvese zvinopa zvakakwana. Niche genres (gamelan, Carnatic, throat kutamba) zvinopesana — ratidza zvakajeka zvinhu zvemumba kuti uwane zvinoshamisa zvibodzwa.
AudioLDM 2 inofamba paGPU yedu yedu yezuva nezuva yemahara pool yekutanga; $ 5 → 200,000 yakabhadharwa ma tokens mushure meizvozvo.
WAV sezvakarongwa (sina kukuvadza, yagadzirira kutambanudza). MP3 iripo musarudzo yemufananidzo yemafaira madiki.
No — AudioLDM 2 inodzosera track yakasviba. Kuti uise muzvidimbu, shandisa /music/separate/ iyo inoshandisa Demucs kuti iite mavocals, drums, bass, uye zvimwe.
Yeah — /batch/ inogamuchira runyorwa rwemaprompts. Each track lands in /account/?tab=history with the prompt as the filename. For more control, use the API.
Yeah — POST yako prompt kuti /v1/music/generate/ with model="AudioLDM 2". /api/ ine yakazara reference kusanganisira lyrics, nguva, uye tempo parameters.
AudioLDM 2 haidzokorore macompositions ane kodzero dzemunyori — ma outputs anogadzirwa nezvinogadzirwa. Free.ai inobvumidza kushandiswa kwekutengesa; iwe unochengeta kodzero dzematracks ako akagadzirwa. Kutaura zita remuimbi muprompt kunobvunza kuti "mumhando ye" — iyo yazvino audio ndiyo yazvino.
15 masekondi kusvika 2 maminitsi patrack zvichienderana nenguva + injini. Premium models inoita zviri nyore. Usati wavhara iyo tab, shandisa iyo queue button — iyo result inoenda ku /account/?tab=history kana yapera.
Yevakurukuru kuti akaenda asi mashoko akanaka, email contact@free.ai neshare link uye isu tichaongorora nyaya-ne-nyaya.