Search AI Products and News

Tencent's X-Omni AI model integrates reinforcement learning to unify image generation and understanding, outperforming rivals in text rendering and visual tasks without classifier guidance.....

The paper by Dr. Liang Wenfeng's team at DeepSeek in collaboration with Peking University has won the ACL 2025 Best Paper Award. This research proposes the Native Sparse Attention (NSA) mechanism, which achieves algorithm-hardware co-optimization through three parallel attention branches: compression, selection, and sliding. It improves the processing speed of long texts by 11 times and outperforms traditional full-attention models. NSA supports a context length of up to 1 million tokens, achieving an 11.6 times decoding speed improvement on 64k text, and a 27B parameter model demonstrates excellent performance on multiple benchmark tests.

Anthropic's AI service Claude has formed a deep collaboration with the design platform Canva, launching a text-to-visual design feature. After users upload written content, the system can automatically identify and generate visual works that match the brand style, offering multiple template choices. This feature significantly lowers the design barrier, helping individual bloggers and corporate marketers quickly create professional visual content and enhance their communication effectiveness. This collaboration marks a new breakthrough in AI within the creative field, and it is expected to bring more innovative design experiences in the future.

iFlytek's upgraded Spark X1 model enhances translation, reasoning, and text generation, achieving top-tier global performance with 20% better translation quality. It scores over 80 in EN-CN conference translation, reduces AI hallucinations, and boosts efficiency for 100+ firms by 50%. Available via web, app, and API.....

On July 23, Kunlun Group introduced two innovative AI models to global users - the Mureka V7 Music Model and the Mureka TTS V1 Audio Model. The launch of these two models marks an important step forward in the field of AI music generation. Since its release, the Mureka V7 Music Model has attracted widespread attention, gaining extensive praise for its efficient music creation capabilities and outstanding music quality. Users simply need to input lyrics, select a style or theme, and click generate to obtain a complete and pleasant song in a short period of time.

The Meta research team introduced an innovative architecture called AU-Net, breaking through the limitations of traditional tokenization techniques. The model learns directly from bytes, dynamically combining bytes into hierarchical sequence representations through a self-regressive U-Net structure. Its contracting path compresses bytes into semantic units, while the expanding path restores information and fuses details, with skip connections preserving local features. A multi-stage pooling strategy is used to extract semantics at the word and phrase levels, and multi-linear upsampling optimizes information fusion. The self-regressive generation mechanism ensures text coherence and significantly improves inference efficiency. This architecture

On July 22, Tencent HuanYuan announced that its self-developed ASR (Speech Recognition) large model has been officially applied to the ima platform, achieving voice input functionality on mobile app interfaces for the first time, offering users a convenient experience of 'speaking fluently'. Users can now directly input questions or record inspirations through voice, without manually typing on the keyboard, greatly improving input efficiency. Tencent HuanYuan's ASR large model stands out with its keen auditory perception and intelligent understanding capabilities, accurately recognizing speech even in complex environments, capable of recognizing 300 words per minute, which is faster than

NVIDIA's Canary-Qwen-2.5B sets a 5.63% WER record on Hugging Face OpenASR. This CC-BY licensed model combines FastConformer encoder with Qwen3-1.7B LLM decoder for efficient speech-to-text and NLP. Supports multi-GPU deployment for cloud/edge applications.....

IndexTTS2 is an upcoming film-grade TTS model with 3 breakthroughs: 1) Local deployment & open weights; 2) Zero-shot voice cloning; 3) Pioneering emotion cloning. Supports precise duration control for dubbing (Chinese/English).....

CASIC introduces Stream-Omni, a multimodal model supporting text, vision, and speech. It uses targeted alignment to reduce data dependency, excels in cross-modal tasks, and offers open-source resources.....

Recently, the French AI laboratory Kyutai announced the official open source of its new text-to-speech model, Kyutai TTS, providing global developers and researchers with a high-performance, low-latency speech synthesis solution. This breakthrough release not only promotes the development of open-source AI technology but also opens up new possibilities for multilingual voice interaction applications. AIbase provides an exclusive analysis of this technological highlight and its potential impact. Ultra-low latency, a new experience in real-time interaction. Kyutai TTS has become an industry standout with its exceptional performance.

El 3 de julio, el instituto francés de investigación en inteligencia artificial Kyutai Labs anunció el lanzamiento del código abierto de su última tecnología de síntesis de voz (TTS), Kyutai TTS, ofreciendo soluciones de generación de voz eficientes y en tiempo real a desarrolladores e entusiastas de la IA. Kyutai TTS destaca por su bajo latencia y sonido de alta fidelidad, permitiendo la transmisión de texto en flujo, comenzando a generar audio sin necesidad de tener todo el texto completo, especialmente adecuado para escenarios de interacción en tiempo real. Kyutai TTS destaca por su excelente rendimiento. Utiliza una sola tarjeta gráfica NVIDIA L40S

The TEN Agent team recently announced that its core models **TEN Voice Activity Detection (VAD)** and **TEN Turn Detection** are now open source, providing powerful technical support for building real-time, multimodal speech AI agents. This move marks a significant advancement in the TEN framework's efforts to promote the democratization and open-source collaboration of speech interaction technology. The following is the latest information compiled by AIbase, offering an in-depth analysis of these two core models.

Recently, a speech synthesis model called Qwen-TTS has made new progress, with its latest version update completed through the Qwen API, bringing users a richer speech synthesis experience. In this update, Qwen-TTS added support for three Chinese dialects: Beijing dialect, Shanghai dialect, and Sichuan dialect, further expanding its application scenarios. The model is trained on a large-scale corpus of more than 3 million hours, achieving naturalness and expressiveness at a human level. Qwen-TTS can not only accurately