Perfect Voice Applications with Chirp and Speech Fine-Tuning

Explore the latest advancements in speech recognition technology with this 27-minute conference talk from Google Cloud Next. Dive into the capabilities of Vertex AI and Chirp, a 2 billion parameter foundation model, revolutionizing speech-to-text applications. Learn how to leverage large models for speech tasks and fine-tune them for specific use cases using in-domain data. Discover the journey from initial setup to perfecting voice models for your applications. Gain insights into the Speech API, Chirp, and Bell Canada's implementation of call listening. Understand the high-level architecture, top 3 use cases, and pitch effectiveness. Explore speech tuning techniques, their implementation, and the impressive results achieved. Perfect for developers and enterprises looking to enhance their voice applications and stay at the forefront of speech recognition technology.

Intro
Speech API
Chirp
Bell Canada
How call listening works
High level architecture
Top 3 use cases
Pitch Effectiveness
Next Steps
Speech Tuning
How we do it
How it works
Results