Discover quality AI tools
Bring together the world’s leading artificial intelligence tools to help improve work efficiency
Articles published by CarlPerez
Alibaba Qwen launched Qwen-Audio-3.1, introducing five models covering speech recognition, synthesis, real-time interaction, and creation. The update slashes prices by 70-95%, making voice tech highly accessible. Key features include context-aware ASR with native transcription polishing, support for 30 languages and 16 Chinese dialects, and ultra-low latency. The TTS-Next model enables general audio generation, combining voices, sound effects, and ambient noise. Real-time interaction supports full-duplex conversation with emotional awareness and tool integration. This comprehensive suite significantly lowers costs while enhancing accuracy, multilingual support, and creative audio production capabilities.
Alibaba Qwen launched Qwen-Audio-3.1, introducing five models covering speech recognition, synthesis, real-time interaction, and creation. The update slashes prices by 70-95%, making voice tech highly accessible. Key features include context-aware ASR with native transcription polishing, support for 30 languages and 16 Chinese dialects, and ultra-low latency. The TTS-Next model enables general audio generation, combining voices, sound effects, and ambient noise. Real-time interaction supports full-duplex conversation with emotional awareness and tool integration. This comprehensive suite significantly lowers costs while enhancing accuracy, multilingual support, and creative audio production capabilities.
Tencent Hunyuan released Hy ASR3.0 preview, a speech recognition model powered by the Hy3 LLM. It improves general recognition, context awareness, multi-scenario robustness, and dialect coverage, evolving from word-by-word transcription to context understanding. On overseas open-source evaluation sets, word error rates reached 3.34% for Mandarin, 2.62% for English, and 3.12% for Cantonese. Tencent's own evaluation set also showed low error rates across various scenarios.
Tencent Hunyuan released Hy ASR3.0 preview, a speech recognition model powered by the Hy3 LLM. It improves general recognition, context awareness, multi-scenario robustness, and dialect coverage, evolving from word-by-word transcription to context understanding. On overseas open-source evaluation sets, word error rates reached 3.34% for Mandarin, 2.62% for English, and 3.12% for Cantonese. Tencent's own evaluation set also showed low error rates across various scenarios.
At Google I/O 2026, the company announced its most significant search overhaul in 25 years, integrating Gemini 3.5 Flash to launch new AI-driven ad formats. These transform ads from passive displays into active conversational services. Key features include an AI Shopping Interpreter that generates personalized purchase rationales, and interactive Shopping Assistants for real-time brand support. Ads are now seamlessly woven into AI summaries and search results, aiming to shorten the path from query to conversion. The new formats, currently in U.S. testing, mark a shift from information matching to AI-facilitated commercial transactions, raising industry discussions about search fairness and user experience.
At Google I/O 2026, the company announced its most significant search overhaul in 25 years, integrating Gemini 3.5 Flash to launch new AI-driven ad formats. These transform ads from passive displays into active conversational services. Key features include an AI Shopping Interpreter that generates personalized purchase rationales, and interactive Shopping Assistants for real-time brand support. Ads are now seamlessly woven into AI summaries and search results, aiming to shorten the path from query to conversion. The new formats, currently in U.S. testing, mark a shift from information matching to AI-facilitated commercial transactions, raising industry discussions about search fairness and user experience.
Despite a combined market cap of $16T and $400B cash, the US Big Five tech firms (Nvidia, Apple, Microsoft, Alphabet, Amazon) disclosed only 10 startup acquisitions in 2025. Alphabet's $32B Wiz deal dominates an otherwise slow M&A trend, with regulatory scrutiny deterring larger moves and talent-poaching emerging as an alternative strategy.
Despite a combined market cap of $16T and $400B cash, the US Big Five tech firms (Nvidia, Apple, Microsoft, Alphabet, Amazon) disclosed only 10 startup acquisitions in 2025. Alphabet's $32B Wiz deal dominates an otherwise slow M&A trend, with regulatory scrutiny deterring larger moves and talent-poaching emerging as an alternative strategy.
Silicon Valley leaders, including Andreessen Horowitz and OpenAI's Greg Brockman, are investing over $100 million in pro-AI PACs to influence 2026 midterms. The 'Leading the Future' network aims to oppose strict AI regulations, advocating for innovation-friendly policies and countering state-level regulatory efforts to maintain U.S. AI competitiveness.
Silicon Valley leaders, including Andreessen Horowitz and OpenAI's Greg Brockman, are investing over $100 million in pro-AI PACs to influence 2026 midterms. The 'Leading the Future' network aims to oppose strict AI regulations, advocating for innovation-friendly policies and countering state-level regulatory efforts to maintain U.S. AI competitiveness.
