Mozilla Hacks: Common Voice dataset tops 20,000 hours. “The latest Common Voice dataset, released today, has achieved a major milestone: More than 20,000 hours of open-source speech data that anyone, anywhere can use. The dataset has nearly doubled in the past year.”
CNN: A Texas teen couldn’t speak with his sister, so he created an app to give her a voice. “Archer Calder spent his childhood unable to talk with his younger sister, Della, his only sibling. Della has Bainbridge-Ropers syndrome, a rare genetic condition which affects her ability to eat, move and speak…. Archer began using his coding abilities to develop a web-based application called Freespeech. The site can be accessed from any device — phones, tablets and computers — with internet access. The simple website lets users program buttons with images of their choice to represent words. When clicked, the word is sounded out loud.”
TechCrunch: Google tests Project Relate, a voice recognition and synthesis app for people with speech impairments. “Google is looking for help developing an Android app aimed at providing more communication options for people with speech impairments. Project Relate, as the effort and app is now called, will provide voice transcription and synthesis that could make it easier for users to be understood.”
Stony Brook University: Fluent: the First Smart Writing Tool for People Who Stutter. “Fluent is the first of its kind. Much of the literature that addresses the intersection of stuttering and AI focuses on a singular facet — stuttering detection. While there is no cure to stuttering, only intervention, this literature looks at diagnosis, not solutions. Fluent addresses the latter. Through the use of AI, the smart writing tool leverages speech patterns of people that stutter, specifically substitution tendencies. Through this, Fluent creates AI-driven inroads for continuous speech.”
New York Times: The Rise and Fall of ‘ZuckTalk’. “ZuckTalk is a style of unpolished speech exhibited in contexts where polish is customary. It’s a linguistic hooded sweatshirt in a metaphorical boardroom. It is more than a collection of tics, but its tics are crucial to understanding it.”
Motherboard: This Tool Lets You Program an Entire App With One Voice Command. “A new tool for generating lines of code from natural speech can churn out programming language from just the sound of your voice. In a video demonstrating the application, Andrew Mayne, who works for OpenAI, says it’s a ‘voice and natural language code creation tool’ that works with Python and Javascript. When you’re done speaking to it, it runs a terminal to test that the code works.”
New York Times: China Punishes Those Who Question ‘Martyrs.’ A Sleuth Keeps Track.. “At least seven people over the past week have been threatened, detained or arrested after casting doubt over the government’s account of the deaths of Chinese soldiers during a clash last year with Indian troops. Three of them are being detained for between seven and 15 days. The other four face criminal charges, including one man who lives outside China…. Their punishment might have gone unnoticed if it weren’t for an online database of speech crimes in China.”
The Next Web: Skyrim modders are using AI to generate new spoken dialogue. “If you’re unimpressed by some of Skyrim’s hilarious dialogue, a new AI app called VASynth lets you take over the scriptwriting. The tool uses voice samples from Bethesda games to convert text into speech. You can generate dialogue in the style of many voices from the publisher’s back catalog, including Skyrim, Fallout 4, and Morrowind.”
PNAS: Speech can produce jet-like transport relevant to asymptomatic spreading of virus. “Medical reports and news sources raise the possibility that flows created during breathing, speaking, laughing, singing, or exercise could be the means by which asymptomatic individuals contribute to spread of the SARS-CoV-2 virus. We use experiments and simulations to quantify how exhaled air is transported in speech. Phonetic characteristics introduce complexity to the airflow dynamics and plosive sounds, such as ‘P,’ produce intense vortical structures that behave like ‘puffs’ and rapidly reach 1 m. However, speech, corresponding to a train of such puffs, creates a conical, turbulent, jet-like flow and easily produces directed transport over 2 m in 30 s of conversation. This work should inform public health guidance for risk reduction and mitigation strategies of airborne pathogen transmission.”
MIT Technology Review: Loud talking could leave coronavirus in the air for up to 14 minutes. “Thousands of droplets from the mouths of people who are talking loudly can stay in the air for between eight and 14 minutes before disappearing, according to a new study. The research, conducted by a team with the US National Institutes of Health and published in PNAS Wednesday, could have significant impact on our understanding of covid-19 transmission.”
Straits Times: National AI database gains 1,000 hours of local English voice samples. “Speech apps and tools such as voice transcription apps could soon be able to pick up Singapore English more accurately, as their developers can now access better data from an expanded corpus of local speech. Some 1,000 hours of natural conversations on topics such as favourite foods and holidays have been added to the National Speech Corpus (NSC), an artificial intelligence (AI) database of locally accented speech maintained by the Infocomm Media Development Authority (IMDA).”
Science Blog: Study: AI Can Detect Depression In A Child’s Speech. “A machine learning algorithm can detect signs of anxiety and depression in the speech patterns of young children, potentially providing a fast and easy way of diagnosing conditions that are difficult to spot and often overlooked in young people, according to new research published in the Journal of Biomedical and Health Informatics.”
Christian Science Monitor: European leaders ask social media companies to censor extremist speech. “The leaders of Britain, France, and Italy will push social media companies on Wednesday to remove ‘terrorist content’ from the internet within one to two hours of it appearing because they say that is the period when most material is spread. British Prime Minister Theresa May, French President Emmanuel Macron, and Italian Prime Minister Paolo Gentiloni will raise the issue at an event on the sidelines of the annual gathering of world leaders at the United Nations.”
Fast Company: Mozilla is crowdsourcing a massive speech-recognition system. “From Amazon’s Alexa to Apple’s Siri, speech recognition and response are becoming mainstays of how we interact with computers, apps, and internet services. But the technology is owned by giant corporations. Now the Mozilla Foundation, maker of the free Firefox browser, is recruiting volunteers to train an open-source speech recognition system.”
TechCrunch: Samsung’s new app uses emojis to help people with language disorders communicate. “Created by Samsung Electronics Italia (the company’s Italian subsidiary) and speech therapist Francesca Polini, Wemogee replaces text phrases with emoji combinations and can be used as a messaging app or in face-to-face interactions. It supports English and Italian and will be available for Android on April 28, with an iOS version slated for future release.”
The ResearchBuzz Firehose is regular RB content presented as individual posts.
For digest-type posts a couple of times a day, please visit ResearchBuzz.
Want to know how to best use the Firehose for content discovery?
Check out this handy article.