Karen Livescu

Title: The Innovative Contributions of Karen Livescu

Introduction

Karen Livescu is a prominent inventor based in Yorktown Heights, NY (US). She has made significant strides in the field of Automatic Speech Recognition (ASR) through her innovative patent. Her work focuses on enhancing the recognition of out-of-vocabulary words, which is crucial for improving the accuracy of speech recognition systems.

Latest Patents

Karen Livescu holds a patent titled "Recognition of out-of-vocabulary in direct acoustics-to-word speech recognition using acoustic word embedding." This patent describes a method, structure, and computer product for learning Out-of-Vocabulary (OOV) words in an ASR system. The invention utilizes an Acoustic Word Embedding Recurrent Neural Network (AWE RNN) to process a character sequence for new OOV words. The RNN generates an Acoustic Word Embedding (AWE) vector, which is then input into an Acoustic Word Embedding-to-Acoustic-to-Word Neural Network (AWE→A2W NN). This network is trained to provide an OOV word weight value from the AWE vector, which is subsequently integrated into the ASR system's existing word embeddings.

Career Highlights

Throughout her career, Karen has worked with notable organizations, including the International Business Machines Corporation (IBM) and the Toyota Technological Institute at Chicago. Her experience in these institutions has contributed to her expertise in speech recognition technologies and machine learning.

Collaborations

Karen has collaborated with esteemed colleagues such as Kartik Audhkhasi and Michael Alan Picheny. Their joint efforts have further advanced the field of speech recognition and have led to innovative solutions in the industry.

Conclusion

Karen Livescu's contributions to the field of Automatic Speech Recognition through her patent and collaborations highlight her role as a leading inventor. Her work continues to influence advancements in speech technology, making significant impacts on how machines understand human language.