Voice big model

On September 19, Mi announced today that Xiaomi-MiMo-Audio, the first original-to-end voice model of open source, had for the first time achieved a generalization of ICL-based samples in the field of voice. According to Mi, five years ago GPT-3 demonstrated for the first time the ability to acquire In-Context Learning (ICL, context learning) through a self-returning language model + large-scale unmarked data training, while in the area of voice, the existing large-scale model still relies heavily on large-scale tagging data and is difficult to adapt to new assignments to human intelligence. And Xia..

September 1, 2011 - Step-Audio today released Step-Audio 2 mini, an open-source end-to-end speech model that has achieved SOTA on multiple international benchmark test sets, and is now available on the Step-Audio 2 mini open platform. Step-Audio 2 mini is now available on Step-Audio Open Platform. 1AI learns from the official introduction that Step-Audio 2 mini models speech understanding, audio reasoning and generation in a unified way, and is the first one to support speech-native Tool Calling capability, which enables networked search and other operations. Step-Audio 2 mini has been tested in several key benchmarks...

Step-1o announced the launch of Step-1o end-to-end speech model with hundreds of billions of parameters, which is said to be "the first end-to-end speech model with hundreds of billions of parameters in China". According to Step-1o, the traditional speech model adopts the cascade program, the user input voice information needs to be converted into text, and then converted into voice output, this process will not only reduce the transmission efficiency, but also in the process of loss of information, including emotions, resulting in the speech model to extend the slow response, the answer to the quality of the level of intelligence is insufficient, the expression of emotion empty stereotypes of the shortcomings. However, an end-to-end speech solution can realize speech understanding and generation of...

In the opening speech of the 2024 vivo developer conference on the morning of October 10, vivo officially released the new blue heart big model matrix, comprehensively upgraded the language big model and end-side big model capabilities, and brought vivo's self-developed voice big model, image big model, and multimodal big model. It is understood that the new blue heart big model matrix includes language big model, end-side big model, speech big model, image big model, and multimodal big model. vivo launched the new 3 billion blue heart end-side big model 3B, which is officially said to be in the dialogue writing, summary summarization, and information extraction ...

On September 5, at the "Creative Boundary and Application Imagination of Big Model" forum of the Bund Conference, Xihu Centron, a big model startup enterprise, formally released and launched the first end-to-end speech big model "Centron Lingo" in China. "Xinchen Lingo" realizes end-to-end speech technology, which directly understands speech, captures the tone, rhythm and emotion when processing conversations, and makes voice replies, which reduces the loss of information processing and makes the "machine" understand people better. As the first end-to-end voice model in China, it creates a new way of human-computer interaction. (CEO of Westlake Centron released the first end-to-end speech grand model in China, Centron L...

In August this year, Westlake Xinchen, invested by Jinke Tomcat, launched the Lingo voice big model, the first end-to-end voice big model in China. The internal beta reservation has been opened today (August 24). In the announcement released on August 21, the official introduction stated that compared with traditional TTS, the end-to-end voice big model is a more comprehensive technology. It not only can recognize speech, but also integrates natural language processing, intent recognition, dialogue management, and speech synthesis, realizing the complete interactive process from speech input to speech feedback, greatly enriching the depth of human-computer interaction…

KU Xunfei's Starfire voice model has received a brand new upgrade, launching a new generation of interactive experience called "Starfire Extreme Hyper Humanoid Interaction". This upgrade has been optimized and enhanced in several aspects, aiming to provide users with a more natural, smooth and emotional dialogue experience. Firstly, the new model achieves faster response time and adopts end-to-end speech-to-speech modeling technology, which enables quick response even in the case of frequent interruptions, and more closely matches the actual situation of daily conversations. Secondly, Starfire Extreme Hyper-Androphilic Interaction has made significant improvements in emotion perception, not only being able to judge the user's emotions based on the speech text...