Whisper · OpenAI
I-OpenAI Whisper iyisikhungo esijwayelekile sokubhala umsindo. Isekela amagama angaphezu kuka-90, ukuhlola okuzenzakalelayo, ukumelana nobuhlakani kanye nokugcizelela.
Umbhalo ongezansi Whisper
I-OpenAI Whisper iyisikhungo esijwayelekile sokubhala umsindo. Isekela amagama angaphezu kuka-90, ukuhlola okuzenzakalelayo, ukumelana nobuhlakani kanye nokugcizelela.
Yini ephawulekayo: Ikwazi ukuqonda amagama angaphezu kuka-90 · Ithobela umsindo kanye nezinhlamvu · Timestamps.
Indlela yokusetshenziswa Whisper
-
1
Ncamashi i-URL yomsindo
Ncamashi i-link eqondile ku-MP3, WAV, MP4 noma noma iyiphi umsindo noma i-video egcinwe ku-server yasekhaya. Iyaxhasa kuze kube amahora ambalwa.
-
2
Ukukhiqizwa
Imodeli ihlela umsindo ngemizuzu engu-30-90 ngokuya ngesikhathi eside.
-
3
Layisha phezulu noma uhlukanise
Layisha phezulu imiphumela noma uhlukanise inxusa eliqondile. Akukho phawu lwamanzi, lilungile ukusetshenziswa.
Imibuzo ebuzwa kaningi
Ingabe udinga ukukhokha imali eningi?
I-Whisper iyinye yezinhlobo ezimahhala ze-catalog. Ukusetshenziswa ngalunye kususa ama-token angama-15 kusuka epulazini lakho, kodwa izinhlobo ezivulekile ezifana ne-Whisper azibizanga, ngakho-ke umkhawulo wezinga uphakeme. I-akhawunti emahhala ivela nge-token engu-500 yokuqala kanye ne-25 engaphezulu ngosuku, ngakho-ke awukwazi ukufinyelela kwikhadi.
Ingabe kukhona umkhawulo wokusetshenziswa kwe-Whisper?
Akukho zindleko zonyaka eziqinile ze-Whisper ku-akhawunti yamahhala: umkhawulo osemthethweni u-rate ngamizuzu/ihora, hhayi ngenyanga. Abangaziwayo bavunyelwe nge-IP; nge-akhawunti ungakwenza okuqukethwe okuningi. Uma ufinyelela ku-500+25 tokens futhi ufuna okuningi, i-Pro $9/month plan ihlanganisa konke.
Yini eyenza i-Whisper ibe hlukile?
OpenAI yilab engezansi kweChatGPT; amamodeli ayo anelwazi olubanzi lwezwe lonke kanye ne-output ehlanzekile. Izinto ezinamandla ezikhethekile zifaka phakathi: Ukuzimela amagama angama-90+ angaphezu kwawo, Ukumelana ne-noise ne-accents, kanye ne-Timestamps.
Ingabe iWisper ithatha isikhathi esingakanani ukudlulisa?
I-Whisper inezinga elilinganiselwe: ~1/5 lesikhathi somsindo ngasinye. Umsindo onamandla noma okhuluma izilimi eziningi ungathatha isikhathi eside.
Ngisebenzisa kanjani i-Whisper ku-ia.gratis?
Ungasebenzisa i-Whisper kusuka ku /audio/transcribe/ ngokufaka ifayela. I-REST API isekela ukufakelwa kwengxenye eziningi ku POST /api/tools/stt/. Isinhlamvu esisheshayo: ukufakelwa okulungile ku 90+ izilimi. Imodeli yangaphakathi yomuntu siqu yi-`whisper`, esetshenziswa uma i-API ixhunywe.