PolySpeech: Exploring Unified Multitask Speech Models for Competitiveness with Single-task Models
This addresses the challenge of building efficient multitask speech models for AI and speech technology applications, but it is incremental as it builds on existing unified model attempts.
The paper tackles the problem of integrating multiple speech processing tasks into a unified model, showing that PolySpeech achieves performance comparable to single-task models across tasks like speech recognition and synthesis.
Recently, there have been attempts to integrate various speech processing tasks into a unified model. However, few previous works directly demonstrated that joint optimization of diverse tasks in multitask speech models has positive influence on the performance of individual tasks. In this paper we present a multitask speech model -- PolySpeech, which supports speech recognition, speech synthesis, and two speech classification tasks. PolySpeech takes multi-modal language model as its core structure and uses semantic representations as speech inputs. We introduce semantic speech embedding tokenization and speech reconstruction methods to PolySpeech, enabling efficient generation of high-quality speech for any given speaker. PolySpeech shows competitiveness across various tasks compared to single-task models. In our experiments, multitask optimization achieves performance comparable to single-task optimization and is especially beneficial for specific tasks.