VibeVoice
Free.ai (self-hosted)
·
tts
·
~500 Token kwa clip
·
4.3
site 3 Ndị ọrụ nke ụdị a
VibeVoice bụ a ngwe-ka-asụsụ olu e mepụtara site na Microsoft. Ọrụ na Long-form audiobooks and multi-speaker podcasts with up to four distinct voices.. Self-hosted na Free.ai GPUs - na-agba ọsọ n'efu megide ụbọchị gị token pool (500 tokens Klipị ọbụla). E wepụtara ya n'okpuru MIT — iji azụmahịa ekwenyela na Free.ai.
Jiri site na API
curl -X POST https://api.free.ai/v1/tts/ \
-H "Authorization: Bearer sk-free-..." \
-H "Content-Type: application/json" \
-d '{"model":"vibevoice","text":"hello world"}'Ajụjụ ndị a jụrụkarị
VibeVoice na-akwado ọtụtụ asụsụ. Ndesịta nkesa ahụ dị n'okpuru engine; form na ihuakwụkwọ a na-anabata ngwe ọbụla na engine ga-egosipụta na asụsụ ndị a na-akwado ya. Gụọ /ọsụsụ/ maka nhọrọ n'ime engine ọbụla ma ọ bụrụ na ịchọrọ asụsụ pụrụ iche.
Ụdị engine ndị kasị ukwuu na-egosipụta n'ụzọ difọ́ọ̀ltụ̀ na-enweghị n'asụsụ Bekee nke America nakwa mpaghara ebe dị mma maka asụsụ ndị na-abụghị Bekee. Ụdị engine Premium nwere ike ịkọwapụta ụdị n'asụsụ Bekee - pịa n'ụdị iji hụ na ọ dị iche.
SSML support varies by engine. Pause, prosody, and emphasis tags are honored on most premium engines and on a few self-hosted ones. Plain text always works — no markup required.
Ntụgharị TTS dị na mbinye engine site na /v1/tts/ API ngwụcha nke na ntụgharị=ezi. UI wéèbụ̀ na ihuakwụkwọ a na-eziga n'ihu n'oge ntụgharị ahụ gachara.
VibeVoice na-arụ ọrụ na GPUs anyị. Ọrụ na-akpụpụta site na pool gị n'ụbọchị n'oge mbụ. Mgbe a na-agwụcha ya, token ndị a na-akwụ ụgwọ na-amalite na $5 → 200,000 token. N'ụzọ dị mfe ~5 token kwa akara, 100 nke ala maka clip.
Ruo 5,000 akara n'otu ajụjụ na webụ UI. Maka ogologo akụkụ (ọdiọnbụị, nsụgharị zuru ezu), jiri /ọdiọnbụị/ọdiọnbụị/ nke na-akpụcha na-akpụcha n'onwe ya, mọọbụ kpọọ API n'ime ntụgharị.
Ya - POST ndesịta nke strings na /v1/tts/batch/, mọọbụ jiri wọkspeesi UI na /ọrụspace/ ka TTS banye n'ime ogologo pipeline (eg, tụgharịa → kwuo → stitch).
Ya - POST ngwe na /v1/tts/ na model="VibeVoice" (mọọbụ slug na ihuakwụkwọ a). Na-eziga WAV mọọbụ MP3. Gụọ /api/ maka nlekọta zuru ezu + snippets SDK.
Ihu a bụ ngwe-ka-asụsụ, ọ bụghị ịsụgharị ụda - ụda bụ difọ́ọ̀ltụ̀ nke engine. Maka ịsụgharị ụda (ịgbanye ụda nlebara anya), gosi /ọsụsụ/ọsụsụ/, nke chọrọ ka ị na-enwe ikike ụda ma ọ bụ na ị nwere ikike edere edere.
Ọrụ ndị na-echekwa onwe ha na-arụ ọrụ na Free.ai-ọbụla GPUs; ọ dịghị ihe na-ahapụ anyị na-arụ ọrụ. Premium engines na-eziga ngwe na-eweta ụdị n'okpuru DPA anyị. Anyị anaghị arụ ọrụ na init gị ma ọ bụ na-ere data.
Ee — Free.ai na-enye ikike iji ozi ọbụla nke ụda a haziri. Laịsịensị nke engine (Apache 2.0, MIT, mọọbụ iwu ndị na-ebubata) a na-egosi n'elu na na ihuakwụkwọ nlebara anya nke móòdù; n'ime usoro ihe omume a pụtara na ọbụna ụda, mgbasaozi, podcasts, nakwa usoroiheomume niile dị n'ime-ọdịiche.
Ee - ọrụ e mehieghị n'aka-ebuli n'aka n'aka n'aka (ọnụọgụgụ ụbọchị ma ọ bụ token ndị a akwụ ụgwọ). Ọ bụrụ na ịkwụghachi na-egosighị n'ụbọchị ahụ, zitere ozi-e contact@free.ai.