# **Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**

Published: 2026-09-23 · Updated: 2026-09-23 · Category: Product Launch

Source page: https://www.sipply.dev/story/know-who-spoke-when-build-real-time-multi-speaker-ai-with-7bb00bc8 · 中文: https://www.sipply.dev/zh/story/know-who-spoke-when-build-real-time-multi-speaker-ai-with-7bb00bc8

## Summary

NVIDIA has unveiled the Nemotron 3 Diarization model, a significant advancement in real-time audio processing technology. This model is designed to handle multi-speaker environments, achieving an impressive Diarization Error Rate (DER) of 14.72% and the capability to distinguish between up to eight speakers. By improving speaker attribution, the Nemotron 3 enhances the quality and utility of transcripts, making it easier for users to understand who spoke when during conversations.

## Key takeaways

- NVIDIA's Nemotron 3 Diarization model offers real-time processing for multi-speaker audio.
- The model achieves a 14.72% Diarization Error Rate, showcasing its accuracy.
- It supports up to eight speakers, enhancing the clarity of audio transcripts.

## Why it matters

This advancement in diarization technology is crucial for applications in various fields such as customer service, media, and education, where understanding speaker contributions is vital. Accurate speaker identification can lead to better insights and more effective communication.

- **Product managers:** For product managers, the introduction of the Nemotron 3 provides an opportunity to enhance products that rely on audio processing, potentially improving user experience and engagement through more accurate transcripts.
- **Developers:** Developers can leverage the capabilities of the Nemotron 3 to build applications that require precise speaker recognition, opening up new possibilities for real-time collaboration tools and voice analytics.
- **Students and job seekers:** Students and job-seekers in fields like data science and AI can benefit from understanding this technology, as it represents a growing area of demand in the job market, particularly in roles focused on audio processing and machine learning.

## Sources

- [**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**](https://huggingface.co/blog/nvidia/nemotron-diarization) (Hugging Face Blog, 2026-09-23)

## 中文

### 了解谁在何时发言：使用NVIDIA Nemotron 3进行实时多说话人AI构建

NVIDIA推出了Nemotron 3 Diarization模型，这是实时音频处理技术的一项重要进展。该模型旨在处理多说话人环境，达到了14.72%的出声错误率（DER），并能够区分多达八位说话者。通过改善说话者归属，Nemotron 3提高了转录的质量和实用性，使用户更容易理解在对话中谁在何时发言。

**要点：**

- NVIDIA的Nemotron 3 Diarization模型提供实时处理多说话人音频的能力。
- 该模型实现了14.72%的出声错误率，展示了其准确性。
- 它支持多达八位说话者，增强了音频转录的清晰度。

**为什么重要：** 这一出声技术的进展对于客户服务、媒体和教育等多个领域的应用至关重要，在这些领域中，理解说话者的贡献是非常重要的。准确的说话者识别可以带来更好的洞察力和更有效的沟通。

---
Sipply: Sip AI News Simply. https://www.sipply.dev