Hong Kong Computer Society

A speech sound disorder (SSD) is a speech disorder in which children cannot produce or use some sounds (phonemes) past the expected age. SSD is the world’s most common type of communication disorder, and it affects approximately 10% to 15% of preschoolers and 6% of students. In the past few decades, researchers focused on how to identify and design the right treatment methods. Machine learning has been widely used in solving problems from various domains. Researchers investigated how to use machine learning approaches in identification, classification, and treatment of SSD in a more effective and efficient way. A survey study on this area can give some new insights to researchers further investigation, so there does not exist a literature review in this area. This survey can fill this gap. In this survey, we discuss the form of data for analysis, methods of data augmentation, adopted feature extraction algorithms, prediction algorithms, performance evaluation techniques. This paper provides a structured view of the research on the use of machine learning in the field of speech sound disorders to date.

Despite the increasing number of species assessed for extinction risk by the International Union for Conservation of Nature (IUCN) (163,040 species as of 2024), only about 1 in 1,000 have been downlisted due to genuine population improvement. Although this rare conservation achievement has been widely celebrated in several recent cases, some other downlisting decisions have met with controversy. A primary role of the IUCN is to assess extinction risk. In this role, it must maintain its independence and not be influenced by the public outcry that may occur when a high‐profile species is downlisted, even if well‐established conservation programs may be disrupted or abandoned as a result. We explored the potential positive and negative consequences of downlisting for conservation efforts through case studies of the giant panda (Ailuropoda melanoleuca), red‐crowned crane (Grus japonensis), saiga antelope (Saiga tatarica), and black‐faced spoonbill (Platalea minor), which has recently been proposed for downlisting. Although downlisting can enable more effective use of limited resources, these cases highlight potential risks, including weakened legal backing, diversion of resources away from the species, and declining public and political support. The relatively unquestioned downlisting of the saiga antelope illustrates how early and inclusive engagement of local experts, assessors, donors, and other stakeholders can help ensure that decisions are effectively communicated and implemented without jeopardizing species recovery. The IUCN Green Status of Species assessment is a complementary tool to the IUCN Red List and offers a useful measure of conservation progress, which can help decision makers ensure that downlisting does not undermine long‐term conservation efforts.