Toward Interactive Dictation

TL;DR

This paper explores enabling natural language spoken commands during voice dictation, introducing a new dataset and system that balances accuracy and latency for real-time speech editing.

Contribution

It presents a novel task, dataset, and approach for open-ended spoken commands in dictation, leveraging large language models for real-time speech segmentation and interpretation.

Findings

Smaller models achieve 30% accuracy with 1.3s latency

Larger models reach 55% accuracy with 7s latency

Demonstrates trade-off between model size, accuracy, and latency

Abstract

Voice dictation is an increasingly important text input modality. Existing systems that allow both dictation and editing-by-voice restrict their command language to flat templates invoked by trigger words. In this work, we study the feasibility of allowing users to interrupt their dictation with spoken editing commands in open-ended natural language. We introduce a new task and dataset, TERTiUS, to experiment with such systems. To support this flexibility in real-time, a system must incrementally segment and classify spans of speech as either dictation or command, and interpret the spans that are commands. We experiment with using large pre-trained language models to predict the edited text, or alternatively, to predict a small text-editing program. Experiments show a natural trade-off between model accuracy and latency: a smaller model achieves 30% end-state accuracy with 1.3 seconds of latency, while a larger model achieves 55% end-state accuracy with 7 seconds of latency.