NVIDIA Speech Artificial Intelligence.

Speech AI Summit. On November 2nd and 3rd day was healed the "online conversations" with experts from Google, Meta, NVIDIA, and more on trends and techniques in automatic speech recognition (ASR) and text-to-speech (TTS) technologies. I participated as a volunteer (Data Scientist & AI - ML Engineer) audience person.

Introduction to Speech AI & End-To-End Speech AI Pipelines & Building Speech AI Applications

Who is this E-Book series for?

This e-book series is intended for business decision owners and developers within an enterprise who would like to understand the core concepts of speech AI and how to build and deploy applications for different use-cases.

The series consists of three parts:

• Part 1: Introduction to Speech AI
• Part 2: End-To-End Speech AI Pipelines
• Part 3: Building Speech AI Applications

Over the course of this e-book series, you will learn about:

• Speech AI and its major building blocks.
• The role of speech AI in industries.
• Different components and technologies that are involved in building end-to-end speech AI pipelines
• How to get started with speech AI in your business with NVIDIA Riva

Overview of Zero-Shot Multi-speaker TTS Systems

Agenda

• Text to Speech (TTS)
• What is zero-shot multi-speaker TTS (ZS-TTS)?
• Multi-speaker TTS versus ZS-TTS
• How important is the Speaker Encoder for ZS-TTS?
• ZS-TTS main papers
• Main ZS-TTS open-source implementations
• ZS-TTS state-of-the-art audio samples

Latest Advances in End -to -End Speech Recognition

Speech representation learning and the emergence of Textless NLP Research

Agenda

• 3 waves of speech representation learning
• SOTA Speech SSL methods
• Generative Spoken LMs
• Privacy-preserving Speech SSL

Unlocking Speech AI Technology for Global Language Users