Namuhla (mhla ziyi-17) ntambama, i-Aliyun imemezele nge-akhawunti yayo esemthethweni ukuthi uchungechunge lwamamodeli ezwi i-Qwen-Audio-3.0 selutholakala ngokusemthethweni epulatifomu ye-Qwen AI kanye nepulatifomu ye-Aliyun Bailian. Abathuthukisi bangabiza amandla amamodeli ngesevisi ye-API, noma babhalisele i-Token Plan ukuze bawasebenzise.
Amamodeli ochungechunge lwe-Qwen-Audio-3.0 angamamodeli ezwi azithuthukisele wona u-Alibaba. Nansi imininingwane yokwakheka komkhiqizo ethulwa yi-IT之家:
Imodeli enkulu yokuqaphela izwi i-Qwen-Audio-3.0-ASR
Imodeli enkulu yokwakha izwi i-Qwen-Audio-3.0-TTS
Imodeli yengxoxo yokuxhumana ngezwi ngesikhathi sangempela i-Qwen-Audio-3.0-Realtime
Ngokwencazelo, ohlwini lwezinga lezwi lwangoJulayi kulo nyaka epulatifomu ehlonishwayo emhlabeni wonke yokuhlola i-AI, i-Artificial Analysis, le modeli ithole indawo yokuqala emhlabeni emikhakheni emithathu: ukuqaphela izwi (ASR), ukuxhumana ngesikhathi sangempela (RealTime) nokwakha izwi (TTS), yathola “i-Grand Slam”.
I-Qwen-Audio-3.0-ASR iguqula izwi libe umbhalo futhi isekela ukuqaphela izimo eziyinkimbinkimbi nezindawo zobungcweti; i-Qwen-Audio-3.0-TTS iguqula umbhalo ube izwi, isekela izilimi eziningi nezilimi zesifunda ezihlukahlukene, futhi ivumela ukulawulwa kwemizwa, iphimbo nesigqi; i-Qwen-Audio-3.0-Realtime isekela ukuqonda nokuxoxa ngezwi kusukela ekuqaleni kuze kube sekugcineni, ikwazi ukulalela nokukhuluma ngesikhathi esisodwa, ukuphazanyiswa noma ukubuzwa imibuzo yokulandelela noma nini, futhi isekela ukubizwa kwamathuluzi.
Njengamanje, amamodeli alolu chungechunge asesetshenziswa emikhiqizweni efana ne-Qwen App, i-Qwen Office kanye ne-Qoder.