François Germain



Recent publications

Numerical models for audio circuits

Single-channel automated speech processing

NMF-based speech and music processing

Spatial Sound

Linear, Local, Learned (L3) camera processing

Other

Recent publications

  • François G. Germain, Kurt J. Werner, “Periodic Analysis of Nonlinear Virtual Analog Models”, In Proceedings of the 2021 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), New Paltz, NY, October 2021. [pdf][bibtex]

  • François G. Germain, Kurt J. Werner, “Practical Virtual Analog Modeling Using Möbius Transforms”, In Proceedings of the 24th International Conference on Digital Audio Effects (DAFx-21), Vienna, Austria, September 2021. [pdf][bibtex]

  • François G. Germain, Qifeng Chen, Vladlen Koltun. “Speech Denoising with Deep Feature Losses”, In Proceedings of Interspeech, Graz, Austria, September 2019. [pdf][github][website][bibtex]

Top

Numerical models for audio circuits

Advanced numerical models

  • François G. Germain, Kurt J. Werner, “Practical Virtual Analog Modeling Using Möbius Transforms”, In Proceedings of the 24th International Conference on Digital Audio Effects (DAFx-21), Vienna, Austria, September 2021. [pdf][bibtex]

  • François G. Germain, “A nonlinear analysis framework for electronic synthesizer effects”, Ph.D. dissertation, Stanford University, Stanford, CA, 2019. [pdf][bibtex]

  • François G. Germain, Kurt J. Werner, “Optimizing Differentiated Discretization for Audio Circuits Beyond Driving Point Transfer Functions”, In Proceedings of the 2017 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), New Paltz, NY, October 2017. [pdf][bibtex]

  • François G. Germain, “Fixed-rate Modeling of Audio Lumped Systems: A Comparison Between Trapezoidal and Implicit Midpoint Methods”, In Proceedings of the 20th International Conference on Digital Audio Effects (DAFx-17), Edinburgh, UK, September 2017. [pdf][bibtex]

  • François G. Germain, Kurt J. Werner, “Joint Parameter Optimization of Differentiated Discretization Schemes for Audio Circuits”, In Proceedings of the 142nd Audio Engineering Society Convention, Berlin, Germany, May 2017. [aes][bibtex]

  • François G. Germain, Kurt J. Werner, “Design Principles for Lumped Model Discretisation Using Möbius Transforms”, In Proceedings of the 18th International Conference on Digital Audio Effects (DAFx-15), Trondheim, Norway, September 2015. [pdf][bibtex]

Virtual analog modeling of audio effects

  • Michael J. Olsen, Kurt J. Werner, François G. Germain, “Network Variable Preserving Step-size Control in Wave Digital Filters”, In Proceedings of the 20th International Conference on Digital Audio Effects (DAFx-17), Edinburgh, UK, September 2017. [pdf][bibtex]

  • Kurt J. Werner, W. Ross Dunkel, and François G. Germain, “A computational model of the Hammond organ vibrato/chorus using wave digital filters”, In Proceedings of the 19th International Conference on Digital Audio Effects (DAFx-16), Brno, Czech Republic, September 2016. [pdf][bibtex]

Nonlinear analysis of vintage audio effects

The interest of researchers and artists for digital implementation of existing vintage audio effects is ongoing. Questions of legacy, and nostalgy explains it, while a lot of these original systems are now rare, and often poorly maintained.

My research focused on blind system identification, by assuming (as it often happens), that the circuitry of the system will not be available in order to perform ad-hoc circuit modeling. The approach is then to fit general-purpose linear model through optimization and machine learning.

  • François G. Germain, Kurt J. Werner, “Periodic Analysis of Nonlinear Virtual Analog Models”, In Proceedings of the 2021 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), New Paltz, NY, October 2021. [pdf][bibtex]

  • François G. Germain, Jonathan S. Abel, Philippe Depalle and Marcelo M. Wanderley, “Uniform Noise Sequences for Nonlinear System Identification”, In Proceedings of the International Conference on Digital Audio Effects (DAFx), York, UK, September 2012. [pdf][bibtex]

  • François G. Germain, “A nonlinear analysis framework for electronic synthesizer effects”, M.A. thesis, McGill University, Montreal, QC, 2011. [pdf][bibtex]

Physical modeling of musical instruments

Satisfactory modeling of the resonant properties of musical instruments has been accomplished with great success during the last decades, in particular using physical modeling.

While challenges remain in order to achieve very expressive sound, most of the limitations of current models reside in two directions: radiation and excitation models. My research has focused on the later, by developing a physically-inspired excitation models for waveguide-based guitar synthesis. The goal being to allow for a realistic control of the synthesis model for artistic usage.

  • François Germain and Gianpaolo Evangelista, “Synthesis of guitar by digital waveguides: Modeling the plectrum in the physical interaction of the player with the instrument”, In Proceedings of the IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), New Paltz, NY, October 2009. [ieee][bibtex]

Top

Single-channel automated speech processing

  • François G. Germain, Qifeng Chen, Vladlen Koltun. “Speech Denoising with Deep Feature Losses”, In Proceedings of Interspeech, Graz, Austria, September 2019. [pdf][github][website][bibtex]

  • François G. Germain, Gautham J. Mysore, Takako Fujioka. “Equalization matching of speech recordings in real-world environments”, In Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Shanghai, China, March 2016. [ieee][pdf][website][bibtex]

  • François G. Germain, Qfeng Chen, Vladlen Koltun, “Speech Denoising with Deep Feature Losses”, U.S. Patent 10,726,858, issued July 28, 2020. [google][bibtex]

  • François G. Germain, Gautham J. Mysore, “Acoustic Matching and Splicing of Sound Tracks”, U.S. Patent 9,601,124, issued March 21, 2017. [google][bibtex]

Top

NMF-based speech and music processing

During the last decade, the power of the representation of audio spectrograms through nonnegative matrix factorization (NMF) has been demonstrated in multiple applications related to speech and music processing.

The recent development of the block-KL NMF algorithm makes it possible to integrate the NMF in a new range of practical applications.

  • François G. Germain, Gautham J. Mysore. “Speaker and Noise Independent Online Single Channel Speech Enhancement”, In Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Brisbane, Australia, April 2015. [pdf][bibtex]

  • François G. Germain, Gautham J. Mysore. “Stopping Criteria for Non-negative Matrix Factorization Based Supervised and Semi-Supervised Source Separation”, Signal Processing Letters, vol.21(10), pp.1284–1288, October 2014. [ieee][pdf][bibtex]

  • Zafar Rafii, François G. Germain, Dennis L. Sun, Gautham J. Mysore, “Combining Modeling of Singing Voice and Background Music for Automatic Separation of Musical Mixtures”, In Proceedings of the International Society of Music Information Retrieval Conference (ISMIR), Curitaba, Brazil, November 2013. [pdf][bibtex]

  • François G. Germain, Dennis L. Sun, Gautham J. Mysore, “Speaker and Noise Independent Voice Activity Detection”, In Proceedings of Interspeech, Lyon, France, August 2013. [pdf][bibtex]

  • François G. Germain, Gautham J. Mysore, “Performance Metric Based Stopping Criteria for Iterative Algorithms”, U.S. Patent 9,866,954, issued January 9, 2018. [google][bibtex]

Top

Spatial Sound

CCRMA has recently acquired a 32-speaker linear array in addition to a spherical 22-speaker array already existing in its Listening Room. These systems are a great opportunity to explore the capabilities of sound field synthesis (Ambisonics, WFS,…) for scientific and artistic purposes.

Top

Linear, Local, Learned (L3) camera processing

  • François G. Germain, Iretayo A. Akinola, Qiyuan Tian, Steven Lansel, Brian A. Wandell. “Efficient Illuminant Correction in the Local, Linear, Learned (L3) Method”, In Proceedings of SPIE/IS&T Electronic Imaging, San Francisco, CA, February 2015. [spie][pdf][bibtex]

Top

Other

Kurt J. Werner, François G. Germain, Cory S. Goldsmith. “Energy-Preserving Time-Varying Schroeder Allpass Filters and Multichannel Extensions”, Journal of the Audio Engineering Society vol.69(7/8), 2021. [pdf][bibtex]

Kurt J. Werner, François G. Germain, “Sinusoidal Parameter Estimation Using Quadratic Interpolation around Power-Scaled Magnitude Spectrum Peaks”, Applied Sciences vol.6(10), 2016. [pdf][bibtex]

Vincent Freour, G. Scavone, Antoine Lefebvre, François Germain. “Acoustical properties of the vocal-tract in trombone performance.” In Proceedings of the 2011 Forum Acusticum, Aalborg, Denmark, 2011. [pdf][bibtex]

Top