Company Filing History:
Years Active: 2007-2008
Title: Daben Liu: Innovator in Audio Classification and Speech Recognition
Introduction
Daben Liu is a notable inventor based in Belmont, MA (US), recognized for his contributions to audio classification and speech recognition technologies. With a total of 2 patents, Liu has made significant strides in enhancing the way audio data is processed and understood.
Latest Patents
Liu's latest patents include innovative systems and methods for classifying audio into broad phoneme classes. This audio classification system is designed to categorize sounds in an audio stream into a limited number of classes, such as vowels, fricatives, narrowband, wideband, coughing, gender, and silence. The system incorporates a signal analysis component and a decoder that utilizes various models for performing these classifications. The classified audio can be instrumental in improving speech recognition capabilities.
Another significant patent by Liu focuses on providing online fast speaker adaptation in speech recognition. This system performs speaker adaptation by receiving an audio segment and determining whether it is a first or subsequent audio segment associated with a speaker turn. The system decodes the first audio segment to generate a transcription and estimates a transformation matrix based on this transcription. For subsequent audio segments, the system uses the transformation matrix to enhance the accuracy of the transcription.
Career Highlights
Throughout his career, Daben Liu has worked with prominent companies such as BBN Technologies LLC and Verizon Corporate Services Group Inc. His experience in these organizations has contributed to his expertise in audio processing and speech recognition technologies.
Collaborations
Liu has collaborated with various professionals in his field, including his coworker Francis G. Kubala. These collaborations have likely enriched his work and led to further advancements in his projects.
Conclusion
Daben Liu's innovative work in audio classification and speech recognition demonstrates his commitment to advancing technology in these areas. His patents reflect a deep understanding of audio processing, which has the potential to significantly enhance user experiences in speech recognition applications.