engblogs

This blog post discusses the work done by Project Euphonia, a team within AI for Social Good at Google Research, to improve automatic speech recognition (ASR) for people with disordered speech patterns such as stuttering, dysarthria, and apraxia. The post explains how they collected speech samples, personalized ASR models, and fine-tuned Google's Universal Speech Model (USM) to better understand and transcribe disordered speech.

Improving the code reading and exploration experiences with new features in GitHub Code View, while maintaining performance

Boost agent productivity by integrating Salesforce for Live Call Analytics, reducing time spent searching for customer information and knowledge articles in various systems.

Deploy Amazon SageMaker Studio lifecycle configurations using the AWS CDK.

Leveraging LLMs for extracting value from unstructured business data

RoboCat is a self-improving robotic agent that learns to operate different robotic arms, solves tasks from as few as 100 demonstrations, and improves from self-generated data.

Introducing a new feature on Roblox that allows people who are 17 and older to discover and engage with age-appropriate experiences.

The blogpost examines Salesforce's cutting-edge security strategy and how they use automation to tackle cyber threats.

This blogpost explores how DoorDash built an ensemble learning model for time series forecasting, focusing on the challenge of balancing accuracy and speed in real-world forecasting applications.

New approaches for detecting AI-generated profile photos in order to improve anti-abuse defenses and protect user experiences on LinkedIn.

Using LLMs to extract value from business data at Snorkel AI's Data-Centric AI Summit in 2022

The blog post discusses two complementary approaches to scene boundary detection in audiovisual content, one using aligned screenplay information and the other using a multimodal sequential model.