Location History:
- Los Altos, CA (US) (2020 - 2023)
- San Mateo, CA (US) (2023)
Company Filing History:
Years Active: 2020-2023
Title: Innovations by Mohammad Shoeybi
Introduction
Mohammad Shoeybi is an accomplished inventor based in Los Altos, California. He has made significant contributions to the field of text-to-speech systems and neural networks. With a total of 3 patents, his work showcases the intersection of technology and innovation.
Latest Patents
One of his latest patents is focused on a real-time neural text-to-speech (TTS) system. This system is constructed from deep neural networks and comprises five major building blocks. These include a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. The segmentation model utilizes deep neural networks with Connectionist Temporal Classification (CTC) loss for phoneme boundary detection. Additionally, the audio synthesis model is a variant of WaveNet that requires fewer parameters and trains faster than the original. This innovative approach allows for faster-than-real-time inference, making the system simpler and more flexible than traditional TTS systems.
Another notable patent involves systems and methods for trace norm regularization and faster inference for embedded models. This patent describes techniques for compressing and speeding up dense matrix multiplications, particularly in fully connected and recurrent layers of neural networks. The trace norm regularization technique leads to improved accuracy and efficiency in training large models. Furthermore, the methods enable faster inference on ARM processors through optimized kernels for small batch sizes.
Career Highlights
Mohammad Shoeybi is currently employed at Baidu USA LLC, where he continues to push the boundaries of innovation in speech recognition and neural networks. His work has garnered attention for its practical applications and advancements in technology.
Collaborations
He has collaborated with notable colleagues such as Sercan Omer Arik and Mike Chrzanowski, contributing to the development of cutting-edge technologies in their field.
Conclusion
Mohammad Shoeybi's contributions to the field of neural networks and text-to-speech systems highlight his innovative spirit and dedication to advancing technology. His patents reflect a commitment to improving efficiency and accuracy in speech recognition, making a significant impact in the industry.