AI services are being implemented using large-scale language models and AI voice technology. NVIDIA is accelerating the creation of an ecosystem by launching related services.

NVIDIA, an AI computing technology company, announced on the 11th that it will hold an NVIDIA AI Developer Meetup online on the 17th with the topics of large-scale language models (LLM) and speech AI.

This meetup will explore global technology trends, solution updates, hands-on demos, and best practices in large-scale language models and voice AI.

Anyone interested in technology can participate in the developer meetup for free, and those who pre-register will receive access to presentation materials and videos, sample code, and live demos.

Large-Scale Language Model (LLM)

Large-scale language models are rapidly growing across a wide range of applications, including healthcare and robotics. In 2021, NVIDIA announced the Megatron 530B large-scale language model. This is a large-scale, customizable language model that can be trained on new domains and languages.

Last September, at GTC, NVIDIA introduced Nemo, a large-scale language model service. This NVIDIA-managed cloud service offers up to a 30% training speedup for large language models as they grow in size and complexity.

Additionally, NVIDIA supports a variety of frameworks for large-scale language model development and inference. Therefore, this developer meetup is expected to provide an opportunity to explore diverse use cases from companies around the world and learn how to implement and infer NVIDIA's large-scale language models.

Riva for Multilingual Voice AI

NVIDIA Riva is a GPU-accelerated software development kit for building and deploying voice AI applications. At GTC last September, NVIDIA announced new features for NVIDIA Riva. Riva is available in multiple languages, including Korean, and features include word-level confidence scores and speaker segmentation (diarization), which identifies speakers in audio streams.

Riva is built to be fully customizable across every stage of the voice AI pipeline, enabling you to efficiently solve unique challenges. Developers can deploy it to their preferred data locations—on-premises, at the edge, or on embedded devices—and businesses can use it to enhance their services, efficiency, and competitiveness. NVIDIA Livaro also allows companies to explore large-scale deep learning models and develop more detailed speech systems.

This developer meetup will provide participants with an introduction to NVIDIA's acceleration solutions for building complex voice AI pipelines and Korean-language voice AI applications.

Meanwhile, this year's AI Developer Meetup will be held online from 2:00 PM to 5:00 PM on the 17th. Registration is available on the registration page .