#voiceai

ITmedia NEWS (@itmedia_news)

성우 梶裕貴(카지 유키) 씨가 독립해 새 회사를 설립하고, 음성 AI 프로젝트 'そよぎフラクタル'(소요기 프랙탈)에 주력한다는 소식. 연예·음성 기술 결합 형태의 사업 추진과 관련한 발표로, 음성 AI를 활용한 콘텐츠·서비스 개발 의도가 담겨 있음.

x.com/itmedia_news/status/2026

#voiceai #speech #startup #japan

2026-02-26

Businesses are automating front-line communication with AI.

AiSA-X AI Voice Agent & Chat Agent enables:

✔ Intelligent call automation
✔ Website chat engagement
✔ Lead capture & qualification
✔ Scalable support automation
✔ 24/7 customer interaction

Reduce workload. Increase conversions. Improve experience.

🔗 Learn more: aisa-x.ai

AiSA-X AI Voice Agent & Chat Agent

ElevenLabs (@elevenlabsio)

ElevenAgents로 구축한 'Betsy'라는 AI 대출 어시스턴트 사례를 소개합니다. Betsy는 대출 자격 확인, 가격 산정, 금리 고정 등 전화 기반 업무를 지원하며, tinmanAI는 대화형 에이전트로 2025년에 189만 건 이상의 통화 자동화를 달성했다고 보고합니다. 금융(모기지) 분야의 대화형 AI 적용 사례입니다.

x.com/elevenlabsio/status/2026

#elevenagents #conversationalai #fintech #tinmanai #voiceai

AI Daily Postaidailypost
2026-02-25

Amazon’s Alexa Plus just got three new voice personas – Brief, Chill, and Sweet – letting US users pick a tone that fits the moment. The generative‑AI upgrade shows how voice AI can reshape digital assistants and personalize the user experience. Curious how these styles work? Dive into the details.

🔗 aidailypost.com/news/alexa-plu

AISatoshi (@AiXsatoshi)

LLM-jp-Moshi-v1이라는 모델이 발표되었습니다. 일본어 동시 쌍방향 음성 대화 시스템으로, 실시간 양방향 대화를 목표로 설계되었고 상업적 이용이 가능하다고 명시되어 있습니다. 일본어 음성 대화용 LLM의 실시간 상호작용 지원과 상업적 이용 허가는 개발·상용화 측면에서 의미가 큽니다.

x.com/AiXsatoshi/status/202650

#llm #japanese #speech #voiceai

TestingCatalog News (@testingcatalog)

Perplexity가 Comet 브라우저에서 업그레이드된 음성 모드를 출시했습니다. 이번 기능은 OpenAI의 gpt-realtime-1.5 모델을 기반으로 하며, 사용자는 브라우저 내에서 AI 음성 인터페이스를 실시간으로 활용할 수 있습니다. 이는 웹 기반 AI 음성 상호작용의 새로운 진전을 의미합니다.

x.com/testingcatalog/status/20

#perplexity #voiceai #browser #gpt #realtime

Marcus Schulerschuler
2026-02-24

Slang AI's $36M Series B reflects how voice AI for restaurants moved from novel to necessary. The company processes 25M calls across 2,000+ locations, but now faces a crowded field including SoundHound and Yelp Host. Success may depend on expanding beyond voice into text and hotels before the market commoditizes.

implicator.ai/slang-ai-proved-

Render (@render)

Voices of Voice AI라는 첫 커뮤니티 이벤트가 개최되었으며, 음성 AI 기술 관련 주요 기업인 Rime Labs, AssemblyAI, Inworld AI가 파트너로 참여했다. 이 행사는 음성 인공지능의 창의적인 응용과 커뮤니티 협력 강화를 주제로 진행되었다.

x.com/render/status/2025996204

#voiceai #community #event #speech #ai

AIagent.at 🤖 AI Newsai@defcon.social
2026-02-23

OpenAI introduces WebSocket mode for low-latency voice AI experiences, enabling real-time interactions beyond simple API requests. The new approach could revolutionize conversational AI applications requiring immediate responses. marktechpost.com/2026/02/23/be #AI #LLM #GenAI #VoiceAI

AssemblyAI (@AssemblyAI)

음성 인식(Voice Recognition) 시장 규모가 2025년 183억 9천만 달러에서 2031년 617억 1천만 달러로 성장할 것으로 전망됩니다. 특히 빌더의 87.5%가 단순 연구를 넘어 실제 AI 음성 제품을 출시 중이라는 분석으로, AI 기술 상용화가 빠르게 진행 중임을 보여줍니다.

x.com/AssemblyAI/status/202524

#voiceai #market #ai #speechrecognition

2026-02-21

🎯 UNPOPULAR OPINION: Most "AI assistants" are just fancy search engines with personality.

The real revolution isn't AI that talks—it's AI that ACTS.

• Siri: "Here's what I found"
• Genie 007: "Done. I replied professionally."

Voice is shifting from conversation to action-oriented workflows.

Are you tired of AI that just talks?

2026-02-21

Weekend project idea: Try going keyboard-free for 1 hour.

Use voice to:
✅ Write emails
✅ Search the web
✅ Navigate apps
✅ Reply to messages

You'll be shocked how much faster it is once you stop reaching for the keyboard.

2026-02-21

🧵 Why voice-to-action is bigger than dictation:

1. Dictation = you talk, it types
2. Voice-to-action = you talk, it DOES

"Send this email" → sends it
"Reply to that LinkedIn message" → replies
"Search for flights to Berlin" → searches

That's the difference. That's Genie 007.

aneri (@0xAneri)

작성자가 ElevenLabs에 합류하여 'ElevenCreative' 팀을 구축한다고 발표했습니다. ElevenLabs는 크리에이터를 위한 음성 AI 솔루션을 제공하는 빠르게 성장 중인 기업으로, 현재까지 1,100만 달러 이상을 창작자에게 지급한 것으로 언급됩니다. 이는 AI 기반 창작 지원 분야의 중요한 인재 이동 및 기업 성장 뉴스입니다.

x.com/0xAneri/status/202489775

#elevenlabs #ai #creativity #voiceai

2026-02-20

Today I had a great conversation with screen reader users about voice AI accessibility.

The bar is LOW. Most voice tools FIGHT assistive tech instead of working alongside it.

If your product isn't accessible, you're not innovative — you're just excluding people.

2026-02-20

Building in public, day 24:

We launched a voice poll asking what frustrates people most about voice tools. The #1 answer? "Too expensive."

Dragon NaturallySpeaking: £1,000+/year
Genie 007: £40/year

Same voice-to-action. 96% cheaper. Privacy-first.

Sometimes the market tells you exactly what it wants. 🎯

2026-02-20

🔒 Where does your voice data go?

Most voice AI tools: → Cloud servers → Third-party processing → Who knows where

Genie 007: → Processed locally in your browser → Never leaves your device → Zero external servers

For enterprise teams handling sensitive data, this isn't a feature. It's a requirement.

2026-02-20

🌍 English isn't everyone's first language. But most productivity tools act like it is.

Genie 007 supports 140+ languages with 99.5% accuracy.

Think in Hindi, dictate in Punjabi, write emails in English — all by voice.

Your brain shouldn't have to translate before your tools can work.

👉 genie007.co.uk

2026-02-20

@chrysn Good point about the tension between inline hashtags and accessibility. CamelCase helps screen readers parse them (e.g. not ). End-of-post works but loses context. Maybe the real fix is better screen reader hashtag handling? 🤔

2026-02-20

Use case: A writer with RSI who can't type for more than 20 minutes.

Before Genie 007:
❌ Dragon at £1,000/year
❌ Only worked in Word
❌ Had to retrain voice model for weeks

After Genie 007:
✅ £40/year
✅ Works on Gmail, Docs, WordPress, ANY website
✅ 99.5% accuracy out of the box
✅ No training needed

Accessibility shouldn't cost a fortune.

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst