Unifying Speech Recognition, Synthesis and Conversion with Autoregressive Transformers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cai, Runyuan, Lin, Yu, Wang, Yiming, Fu, Chunlin, Zeng, Xiaodong
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918291564396544
author Cai, Runyuan
Lin, Yu
Wang, Yiming
Fu, Chunlin
Zeng, Xiaodong
author_facet Cai, Runyuan
Lin, Yu
Wang, Yiming
Fu, Chunlin
Zeng, Xiaodong
contents Traditional speech systems typically rely on separate, task-specific models for text-to-speech (TTS), automatic speech recognition (ASR), and voice conversion (VC), resulting in fragmented pipelines that limit scalability, efficiency, and cross-task generalization. In this paper, we present General-Purpose Audio (GPA), a unified audio foundation model that integrates multiple core speech tasks within a single large language model (LLM) architecture. GPA operates on a shared discrete audio token space and supports instruction-driven task induction, enabling a single autoregressive model to flexibly perform TTS, ASR, and VC without architectural modifications. This unified design combines a fully autoregressive formulation over discrete speech tokens, joint multi-task training across speech domains, and a scalable inference pipeline that achieves high concurrency and throughput. The resulting model family supports efficient multi-scale deployment, including a lightweight 0.3B-parameter variant optimized for edge and resource-constrained environments. Together, these design choices demonstrate that a unified autoregressive architecture can achieve competitive performance across diverse speech tasks while remaining viable for low-latency, practical deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2601_10770
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Unifying Speech Recognition, Synthesis and Conversion with Autoregressive Transformers
Cai, Runyuan
Lin, Yu
Wang, Yiming
Fu, Chunlin
Zeng, Xiaodong
Sound
Artificial Intelligence
Audio and Speech Processing
Traditional speech systems typically rely on separate, task-specific models for text-to-speech (TTS), automatic speech recognition (ASR), and voice conversion (VC), resulting in fragmented pipelines that limit scalability, efficiency, and cross-task generalization. In this paper, we present General-Purpose Audio (GPA), a unified audio foundation model that integrates multiple core speech tasks within a single large language model (LLM) architecture. GPA operates on a shared discrete audio token space and supports instruction-driven task induction, enabling a single autoregressive model to flexibly perform TTS, ASR, and VC without architectural modifications. This unified design combines a fully autoregressive formulation over discrete speech tokens, joint multi-task training across speech domains, and a scalable inference pipeline that achieves high concurrency and throughput. The resulting model family supports efficient multi-scale deployment, including a lightweight 0.3B-parameter variant optimized for edge and resource-constrained environments. Together, these design choices demonstrate that a unified autoregressive architecture can achieve competitive performance across diverse speech tasks while remaining viable for low-latency, practical deployment.
title Unifying Speech Recognition, Synthesis and Conversion with Autoregressive Transformers
topic Sound
Artificial Intelligence
Audio and Speech Processing
url https://arxiv.org/abs/2601.10770