New open voice model listens continuously and decides every 0.4 seconds if necessary by

A new open voice model fits perfectly into your environment by listening constantly. It is not GPT-4o or Qwen3.5-Omni who waits for a piece to say. Audio Interaction does not drop any tracks before evaluating the text using its template.

This new system was designed to be as flexible as possible. His model can translate, transcribe and exchange conversations while detecting daily noises such as cough during a conversation. Technical details are quite simple to understand, but it is important to note that this is an open source system.

Source code, model weight and installation instructions are available on GitHub under the Apache 2.0 license. Training materials for learning were also provided. It is a tool that could be particularly useful in areas such as noise detection or voice recognition.

This article, "New Voice Model Listens Everywhere", was published on The Decoder to inform you about the new open voice system. This type of evolution continues to improve our understanding and interaction with natural language.

---

*Source : [The Decoder](https://the-decoder.com/new-open-source-voice-model-listens-nonstop-and-decides-every-0-4-seconds-whether-to-speak-or-stay-silent/). Synthèse éditoriale assistée par IA — TechnoExpress.*