Shahab Jalalvand

Title: Shahab Jalalvand and Conversation-Aware Speech Recognition.

Introduction

Shahab Jalalvand is an inventor associated with Interactions Corporation and based in Summit, New Jersey. His patent work includes a system for using neural networks and hierarchical language models to transcribe conversation audio with awareness of dialogue context.

Latest Patents

Jalalvand is credited with 1 patent: “Neural Network Based Conversation-Aware Automatic Speech Recognition.” The invention processes audio inputs representing utterances in a conversation and determines a dialogue state for each utterance. It uses a hierarchical language model with a top-level model trained on sequences of dialogue states and lower-level models trained on utterances associated with each state. The system then applies the trained model to transcribe audio from new conversations.

Career Highlights

Jalalvand’s listed patent reflects work on applying machine learning to automatic speech recognition. By incorporating dialogue states into the transcription process, the system is designed to use conversational context when interpreting spoken language. He is associated with Interactions Corporation.

Collaborations

The available information connects Jalalvand’s patent and professional affiliation with Interactions Corporation. No additional collaborators are specified in the provided details.

Conclusion

Shahab Jalalvand’s patent describes a context-aware approach to speech recognition that combines neural-network-based language modeling with dialogue structure to transcribe conversations.