VITA-Audio: Fast Interleaved Cross-Modal Token Generation for Efficient Large Speech-Language Model

TL;DR

VITA-Audio introduces a fast, multi-modal large speech model that significantly reduces latency in audio token generation, enabling real-time speech applications with 3-5x faster inference and strong benchmark performance.

Contribution

The paper presents a novel multi-modal large speech model with a lightweight token prediction module and a progressive training strategy for real-time audio generation.

Findings

Achieves 3-5x inference speedup at 7B parameters.

Outperforms similar models on ASR, TTS, and SQA benchmarks.

Enables real-time conversational speech applications.

Abstract

With the growing requirement for natural human-computer interaction, speech-based systems receive increasing attention as speech is one of the most common forms of daily communication. However, the existing speech models still experience high latency when generating the first audio token during streaming, which poses a significant bottleneck for deployment. To address this issue, we propose VITA-Audio, an end-to-end large speech model with fast audio-text token generation. Specifically, we introduce a lightweight Multiple Cross-modal Token Prediction (MCTP) module that efficiently generates multiple audio tokens within a single model forward pass, which not only accelerates the inference but also significantly reduces the latency for generating the first audio in streaming scenarios. In addition, a four-stage progressive training strategy is explored to achieve model acceleration with minimal loss of speech quality. To our knowledge, VITA-Audio is the first multi-modal large language model capable of generating audio output during the first forward pass, enabling real-time conversational capabilities with minimal latency. VITA-Audio is fully reproducible and is trained on open-source data only. Experimental results demonstrate that our model achieves an inference speedup of 3~5x at the 7B parameter scale, but also significantly outperforms open-source models of similar model size on multiple benchmarks for automatic speech recognition (ASR), text-to-speech (TTS), and spoken question answering (SQA) tasks.