Scaling Properties of Speech Language Models
This work addresses the challenge of scaling SLMs for better syntax and semantics, which is incremental as it builds on existing scaling theories.
The paper investigates the scaling properties of Speech Language Models (SLMs) to estimate the compute needed for SLMs to match text-based LLMs in English proficiency, finding that SLMs scale up to three orders of magnitude more slowly than LLMs.
Speech Language Models (SLMs) aim to learn language from raw audio, without textual resources. Despite significant advances, our current models exhibit weak syntax and semantic abilities. However, if the scaling properties of neural language models hold for the speech modality, these abilities will improve as the amount of compute used for training increases. In this paper, we use models of this scaling behavior to estimate the scale at which our current methods will yield a SLM with the English proficiency of text-based Large Language Models (LLMs). We establish a strong correlation between pre-training loss and downstream syntactic and semantic performance in SLMs and LLMs, which results in predictable scaling of linguistic performance. We show that the linguistic performance of SLMs scales up to three orders of magnitude more slowly than that of text-based LLMs. Additionally, we study the benefits of synthetic data designed to boost semantic understanding and the effects of coarser speech tokenization.