{"@context":"https://neupai.io/schema/v0.2","@type":"StructuredNewsArticle","identity":{"article_id":"tech42_20261022_voice-ai-turn-detection-competition","canonical_url":"https://www.tech42.co.kr/%eb%82%b4-%eb%a7%90-%ec%9e%90%eb%a5%b4%ec%a7%80-%eb%a7%88%ec%9d%8c%ec%84%b1-ai-%ea%b2%bd%ec%9f%81-%ec%96%bc%eb%a7%88%eb%82%98-%eb%b9%a8%eb%a6%ac%eb%b3%b4%eb%8b%a4-%ec%96%b8/?utm_source=rss&utm_medium=rss&utm_campaign=%25eb%2582%25b4-%25eb%25a7%2590-%25ec%259e%2590%25eb%25a5%25b4%25ec%25a7%2580-%25eb%25a7%2588%25ec%259d%258c%25ec%2584%25b1-ai-%25ea%25b2%25bd%25ec%259f%2581-%25ec%2596%25bc%25eb%25a7%2588%25eb%2582%2598-%25eb%25b9%25a8%25eb%25a6%25ac%25eb%25b3%25b4%25eb%258b%25a4-%25ec%2596%25b8","ai_url":null,"publisher":{"name":"테크42","domain":"www.tech42.co.kr","type":"online"},"author":"정재엽 기자","published_at":"2026-10-22T00:06:23.000Z","updated_at":null,"language":"en","article_type":"straight_news","originality":"self_produced"},"content":{"headline":"Don't Interrupt Me…In the Voice AI Race, It's Not 'How Fast' But 'When to Speak'","summary":"The core competition in voice AI is shifting from response speed to end-of-speech detection (turn detection) technology. Mistral AI, ElevenLabs, Channel Corporation and others have successively released new models emphasizing natural conversation flow and avoiding interruptions.","topics":["voice ai","artificial intelligence","turn detection","customer service"],"geography":["KR","FR","US"],"entities":[{"name":"Mistral AI","canonical_id":"corp:fr:mistral-ai","type":"company","role_in_article":"primary_subject","metadata":{"ticker":null,"parent":null}},{"name":"ElevenLabs","canonical_id":"corp:us:elevenlabs","type":"company","role_in_article":"primary_subject","metadata":{"ticker":null,"parent":null}},{"name":"Channel Corporation","canonical_id":"corp:kr:channel-corporation","type":"company","role_in_article":"primary_subject","metadata":{"ticker":null,"parent":null}},{"name":"Rincle","canonical_id":"corp:kr:linkle","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}}],"claims":[{"id":"c1","statement":"Mistral AI's voice AI model 'Voxtral 3' supports speech recognition in 50 languages, including Korean","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"October 20 (local time)","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":50,"unit":"languages","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c2","statement":"Mistral AI explained that a single exchange of listening, thinking, and responding is completed within 1 second","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"October 20 (local time)","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":1,"unit":"seconds","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c3","statement":"ElevenLabs' 'Eleven Realtime' model supports more than 90 languages","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"the same day","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":90,"unit":"languages","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c4","statement":"ElevenLabs' 'Max' model begins speaking 0.57 seconds after receiving a question","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"the same day","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":0.57,"unit":"seconds","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c5","statement":"ElevenLabs' 'Mini' model scored 95.8% in conversational naturalness evaluation, surpassing 'Max' (94.1%)","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"the same day","source_type":"company_disclosure","comparison":"internal_model_comparison","type":"fact","figures":{"value":95.8,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c6","statement":"Channel Corporation analyzed about 10,000 AI phone orders at an in-house cafe and found that in 87% of cases where the AI waited, customers continued speaking within 1.5 seconds","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"the 21st","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":87,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c7","statement":"Food waste disposer brand Rincle said that after introducing 'Voice Alf,' the number of calls directly answered by human agents dropped by about 39%","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"the 21st","source_type":"company_disclosure","comparison":"pre_implementation","type":"fact","figures":{"value":39,"unit":"%","approximate":true,"converted":null},"expiry_hint":null,"insight":null},{"id":"c8","statement":"Food waste disposer brand Rincle said that after introducing 'Voice Alf,' call duration decreased by about 42%","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"the 21st","source_type":"company_disclosure","comparison":"pre_implementation","type":"fact","figures":{"value":42,"unit":"%","approximate":true,"converted":null},"expiry_hint":null,"insight":null},{"id":"c9","statement":"ElevenLabs was valued at approximately ₩29.15 trillion ($22 billion) in late September","as_of":"2026-09","as_of_explicit":true,"as_of_raw":"late September","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":29150000000000,"unit":"KRW","approximate":true,"converted":{"value":22000000000,"unit":"USD"}},"expiry_hint":null,"insight":null},{"id":"c10","statement":"ElevenLabs stated that its AI processes more than 15 million conversations per week","as_of":"2026-10","as_of_explicit":false,"as_of_raw":"October 2026","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":15000000,"unit":"conversations/week","approximate":false,"converted":null},"expiry_hint":null,"insight":null}],"ai_emotional_context":{"valence":0.2,"arousal":0.3,"primary_emotions":[{"emotion":"optimistic","intensity":0.5}],"secondary_emotions":[{"emotion":"curious","intensity":0.3}],"emotional_triggers":[{"claim_id":"c6","emotion":"optimistic","reason":"significant_positive_metric"}]},"image":{"url":"https://onjblseywainslkhvkav.supabase.co/storage/v1/object/public/article-images/504951f9-4189-4ef9-bdd4-634d5bc13e89.png","alt":"전화 상담 중 AI 챗봇과 소통하며 답답해하는 여성의 모습을 묘사한 일러스트, 음성 인식 오류와 소통 지연을 표현.","caption":"(AI 생성 이미지)","source":"first_img","alt_status":"auto"}},"provenance":{"source_chain":["primary_reporting"],"original_source_url":null,"related_articles":[]},"temporal":{"freshness":"recent","next_update_expected":null},"access":{"license":"neupai_standard","attribution_required":true,"structured_data":"free","full_text_available":false,"full_text_access":null}}