Brian Kingsbury

Timeline

1. 1999

   Joining IBM as a researcher

   Before joining IBM, he earns an undergraduate degree in electrical engineering at Michigan State University and pursues a computer-science PhD at the International Computer Science Institute and UC Berkeley. He joins IBM as a researcher in 1999.

2. By 2010

   The modulation-filtered spectrogram

   At UC Berkeley's Computer Science Division and ICSI's Realization Group, his PhD research under Nelson Morgan develops the modulation-filtered spectrogram (MSG), a front end for robust speech recognition in reverberant environments. He presents the talk "Machine Recognition of Speech" in 2010.

3. 2012

   Deep neural speech models

   In 2012, he coauthors an overview of deep neural networks for acoustic modeling in speech recognition; it reports that the models outperform Gaussian mixture models across speech-recognition benchmarks.

4. 2012

   Distributed Hessian-free training

   With Tara N. Sainath and Hagen Soltau, he coauthors a paper on distributed Hessian-free optimization for training deep neural acoustic models in 2012. On a 50-hour broadcast-news task, the method yields a 4.4% relative improvement in word error rate.

5. 2018

   IEEE Fellow

   In 2018, IEEE elevates him to Fellow for contributions to acoustic models for automatic speech recognition.

6. 2025

   Granite-speech models

   In 2025, he coauthors the Granite-speech paper, which describes compact speech-language models for English automatic speech recognition and translation. It specifies 2B- and 8B-parameter variants that are available for research and commercial use under Apache 2.0.

7. September 2026

   Granite 5.0 TurboCTC

   In September 2026, he coauthors a paper describing Granite 5.0 TurboCTC, a 470-million-parameter encoder-only ASR model. The paper reports English short-form speech recognition performance on the Open ASR leaderboard and a speed twice that of its fastest competitor.

8. As of October 2026

   Distinguished research scientist

   As of October 2026, he is a distinguished research scientist at IBM Research in Yorktown Heights and studies machine learning and large-vocabulary speech recognition, with a focus on acoustic modeling and robustness. He serves as senior area editor for IEEE/ACM Transactions on Audio, Speech, and Language Processing and action editor for Transactions of Machine Learning Research and Transactions of the Association for Computational Linguistics.