Company Filing History:
Years Active: 2025
Title: Ekaterina Petrova: Innovator in Audio Generation Models
Introduction
Ekaterina Petrova is a prominent inventor based in Oberhaching, Germany. She has made significant contributions to the field of audio generation, particularly in enhancing datasets for training models that synthesize speech. Her innovative work is paving the way for advancements in text-to-speech technology.
Latest Patents
Ekaterina holds a patent titled "Augmenting datasets for training audio generation models." This patent focuses on augmenting a target voice dataset using speech prediction techniques. The invention involves training encoder and decoder models to encode audio data into encoded speech data and convert it back to audio. The encoded units include semantic information, such as phonemes and words, along with feature data that indicates prosody, timbre, speaker identity, speech style, and emotion. An acoustic/semantic language model (ASLM) is configured to predict encoded speech data similarly to how a language model predicts words. This technology allows for the generation of synthesized speech samples that closely resemble the characteristics of the target voice dataset. The augmented dataset significantly improves the performance of text-to-speech models compared to training with only the original dataset. Ekaterina has 1 patent to her name.
Career Highlights
Ekaterina is currently employed at Amazon Technologies, Inc., where she continues to innovate in the field of audio technology. Her work is instrumental in developing advanced solutions that enhance user experiences through improved speech synthesis.
Collaborations
She collaborates with talented individuals such as Mateusz Aleksander Lajszczak and Adam Marek Gabrys, contributing to a dynamic and innovative work environment.
Conclusion
Ekaterina Petrova's contributions to audio generation models exemplify her commitment to innovation in technology. Her work not only advances the field but also enhances the capabilities of text-to-speech systems, making a lasting impact on how we interact with technology.