Zillennials Use Speech More Often

Plus: A new speech framework improves medical outcomes.

Researchers Expand on-Device Speech Recognition

New algorithm eliminates the need for an internet connection and lots of RAM.

Not-So-Natural Language

Users tend to anthropomorphize speech systems in sometimes surprising ways.

Voice’s Disease Detection Capabilities Expand

New research has expanded the use cases and the types of medical conditions that voice biomarker technology can address.

Microsoft Turbocharges Voice Capabilities in Windows 11

The software giant is hard at work bringing voice features to more PCs.

Researchers Have Built an AI-Powered Sarcasm Detector

Researchers in the Netherlands used American sitcoms  to train a neural network.

Amazon Proposes the SpeechVerse Framework

SpeechVerse integrates textual LLMs with speech encoders in one supervised training setup for a more comprehensive understanding of both speech and text.

Apple Proposes Acoustic Model Fusion to Improve Speech Recognition

Apple has proposed a new approach that integrates external acoustic models into end-to-end ASR systems, aiming to refine speech recognition by enriching systems with broader acoustic knowledge.

Voice’s New Frontier: Diabetes Detection

Research has found significant vocal variations in people with and without type 2 diabetes and suggests that voice technologies could be used to detect the condition.

NYC Mayor Uses Voice Tech for Multilingual Robocalls

The robocalls, delivered via voice clone tech, present information on city job fairs and other events in Spanish, Yiddish, Haitian Creole, Mandarin and Cantonese Chinese, and several other languages.

Meta, Google Develop Their Own AI Speech Models

Facebook parent company Meta last month unveiled Voicebox, an advanced AI tool for generating speech from text, while Google introduced AudioPaLM, its own large language model that can tackle speech understanding and generation tasks.

Lawmakers Propose a Ban on Government Use of Biometrics

Carnegie Mellon ASR Pipeline Seeks to Recognize 1,900 Languages Without Audio

A research team from Carnegie Mellon University in Pittsburgh created a voice recognition pipeline that does not need audio to model low-resource languages.

Stanford Model Teaches Turn-Taking to Virtual Assistants

With the goal of creating a more natural conversational flow, a team of researchers at Stanford University replaced the classification approach traditionally used with a more continuous approach.

European Union Reins in Big Tech

The Digital Services Package will prohibit so-called gatekeepers from restricting consumers' access to third-party voice technologies and other applications and digital services.

Microsoft Reins in Its Voice Capabilities

The tech giant said its measures would amount to a significant update to the "Responsible AI Standard" it first released in 2001.

Speech Can Help Students Read, Experts Conclude

Voice-driven tools have had a direct impact on fluency and literacy, unlocking far more natural and nuanced ways of gauging reading fluency and comprehension than traditional quizzes and tests.

Voice Soars as Vehicle Interface of Choice

In-car voice controls must be context-based to offer the best user experience.

Voicebots Outperformed Humans in COVID Detection

Leveraging VOIQ's artificial intelligence-powered voice software to have conversations with citizens over the phone, Colombia became the first country globally to use voicebots at this scale to diagnose millions of individuals within a matter of hours vs. weeks or months with a solely human-powered team.

Overheard/Underheard: A New Collar Gives Dogs a Voice

An AI-powered dog collar uses voice recognition technology to detect and analyze five emotional states—happy, anxious, angry, sad, or relaxed—in dogs and tracks their physical activity.