Nhazi ụda
Ọrụ ọhaneze OK
567 models
Enweghị akara mmiri
Enweghị mkpa ịbanye
Móòdù:
+ GPT-5, Claude, Gemini
Na-ekewapụta ụda n'ime stiimi...
O nwere ike ịtụnye 30-60 sekọndSteelụ ndị ahụ e wepụrụ
Nkarachọ ndị ahụ
Nke ahụ
Token na-aga n'okpuru.
Wepụta Token
❤️ Ịhụnanya Free.ai? Kpọtụrụ enyi gị!
Sign up iji nweta a referral njikọ na-enweta 30,000 tokens kwa enyi.
Ịchọrọ ihe ọzọ?
Nweta n'efu: 30,000 tokens/ụbọchị
Ndebanye aha
Na-arụ ọrụ n'ihe nchọgharị gị...
Split any song into 4 stems with AI: vocals, drums, bass, other. MP3 / WAV in, ZIP out. Open-source Mel-Band RoFormer + SCNet, fully free.
Otu esi eji ya Nhazi ụda
1
Tinye inu gị
Tinye ngwe, bubata faịlụ, mọọbụ gosi ihe ịchọrọ. Achọrọghị akaụntụ.
2
Pịa iji mepụta
Anyị AI na-enyocha arịrịọ gị n'ime sekọnd na-eji ihe kacha mma open-source models.
3
Bubata na akwado
Bubata, debata, mọọbụ kesaa nsonaazụ gị. Free maka ojiji nkeonwe na nke azụmahịa.
Jiri tùlè a site na API
Megharịa ihenhọrọ a site na koodị gị. OpenAI-compatible REST endpoint, Bearer-token auth, enweghị SDK ọzọ achọrọ. Token costs match the web interface.
curl -X POST https://api.free.ai/v1/music/generate/ \
-H "Authorization: Bearer sk-free-..." \
-H "Content-Type: application/json" \
-d '{"prompt": "upbeat synthwave instrumental", "duration": 30}'Nhazi ụda - FAQ
Ọ na-ewepụ ọbụla n'ime 4 stims na AI: vocals, drums, bass, na ndị ọzọ (synths/piano/guitar/strings). Stims ọbụla bụ faịlụ WAV ya. Vocals na-abịa site na open-source Mel-Band RoFormer model; drums, bass na ndị ọzọ na-abịa site na open-source SCNet model.
Vokal-ewepụtụ na-echekwa ihe niile n'agbanyeghị vocals (instrumental output). Isolaịtị vocals na-eme ihe n'okporo ụzọ (acapella output). Ọrụ n'otu n'otu na-enye gị steelụ anọ niile n'otu n'otu ka ị nwee ike ịkpọkọta na ịkọ. Oge n'otu n'ime n'okpuru hoodu.
A ZIP archive na 4 WAV faịlụ: vocals.wav, drums.wav, bass.wav, other.wav. Otu ọ bụla bụ ogologo na ọnụọgụgụ saịmpọn dị ka init, ya mere ha na-echekwa n'ụzọ dị ọcha mgbe a na-atụgharị ha n'ime DAW.
MP3, WAV, M4A, FLAC, OGG, AIFF, na ndị kasị ewu ewu audio formats. Max ibudata 50 MB (ihe banyere 5 nkeji nke MP3).
6-15 sekọnd kwa nkeji nke input na A100 anyị. A 3-minute song na-ewe banyere 25-40 sekọnd.
Ọ dị mma n'ihe nchọgharị redio na-emegharị n'ụzọ ziri ezi - vocals bụ ihe dị ka zuru ezu, drums na bass dị ọcha, "ọnụọgụgụ" mgbe ụfọdụ nwere obere crosstalk site na stiimi ndịda. Lo-fi recordings, live shows, or heavily-effected tracks produce noisier separations.
Four stems (vocals, drums, bass, other) is the split the open-source models we run are trained for, and it is where their quality is strongest. We do not offer a 6-stem split with separate guitar and piano.
Yabụ - ihuakwụkwọ nsonaazụ ahụ na-egosi bọtịnụ mbipụta nkeonwe maka vocals.wav, drums.wav, bass.wav, other.wav n'akụkụ ZIP niile-n'otu.
Ọrụ onwe onye (ụdị, ọmụmụ, mmegharị maka nhụjuanya na-enweghị n'aka) bụ n'ozuzu ya n'ụzọ ziri ezi. Nhazi, mbipụta, mọọbụ n'ụzọ na-ahapụ n'aka na-achọ sync / mechanical / sample license nke ọbụna. I nwere ikike.
N'ihe dịka 3,000 token n'otu n'otu, flat n'otu n'otu-ịjụjụ (n'enweghị n'ihe banyere ogologo nke n'otu n'otu, ebe ọ bụ na separator na-arụ otu oge n'otu n'otu n'ime ụda ahụ). Free daily-pool tokens cover most use.
Spleeter bụ ihe ochie nke na-emeghe isi mmalite. Mel-Band RoFormer na SCNet architectures anyị na-arụ ọrụ bụ ndị ọhụrụ na-emeri nke ọma n'elu ya na MUSDB18 benchmark nke ndị mmadụ; na 50 MUSDB18 test extracts anyị vocal stem measures 11.3 dB SDR. Moises bụ subscription na-akwụ ụgwọ, ebe Free.ai bụ ụgwọ-n'oge iji na-enweghị n'efu ụbọchị pool.
Ya - POST ụda na /v1/ọdị/separate/ na `mode=stems` iji nweta stems niile n'ihu. Mọọbụ kọwaa `mode=vocals` / `mode=instrumental` maka ụdị-otu-stem. Doktọ na /api/.
Olee otú ị ga-esi họrọ ihenhọrọ a?