It appears that French AI startup Kyutai has created a new chatbot named "Moshi" using AI, which is modeled off GPT-4o. Moshi is a voice-activated artificial intelligence assistant that Kyutai, a French artificial intelligence firm, reportedly unveiled. It appears that Kyutai Labs introduced a fresh take on AI chatbots following the negative reception that OpenAI's chatGPT received. It makes sense that Moshi would be positioned as a competitor to OpenAI's GPT-4o. According to early reports, many users did not find the GPT-4o's "voice mode" to be satisfactory. In the middle of this, Kyutai's Moshi AI chatbot appears to offer a better "voice mode."

(Source: Google Images) 

According to Kyutai, Moshi speaks with a variety of accents. Furthermore, it is anticipated that Moshi will possess approximately seventy distinct emotional and speech patterns. Even two audio streams can be handled simultaneously by the AI. This implies that you should anticipate an AI chatbot that converses "human-like."


(Source: Google Images) 

Additional Updates ~

Kyutai appears to have stated that over 100,000 artificial dialogues created with Text-to-Speech (TTS) technology were refined during the creation of Moshi. Moreover, Kyutai intends to assist Moshi in learning the subtleties and tones of human communication. It's also said that the brand worked with a professional voice actor to improve Moshi's vocals. Moshi is intended to facilitate realistic voice discussions with consumers, like to those offered by Google Assistant or Alexa. Nevertheless, the Helium 7B model powers Moshi. In a video demonstration, Kyutai demonstrated Moshi's talents. The Kyutai team engaged with Moshi throughout the presentation to show it as a mentor or friend. How the display embodied characters in roleplays further demonstrated its inventiveness.