TASTE-Streaming: Towards Streamable Text-Aligned Speech Tokenization and Embedding for Spoken Language Modeling

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Tseng, Liang-Hsuan, Lee, Hung-yi
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910050821341184
author Tseng, Liang-Hsuan
Lee, Hung-yi
author_facet Tseng, Liang-Hsuan
Lee, Hung-yi
contents Text-speech joint spoken language modeling (SLM) aims at natural and intelligent speech-based interactions, but developing such a system may suffer from modality mismatch: speech unit sequences are much longer than text tokens. Prior work reduces this gap with text-aligned tokenization and embedding (TASTE), producing speech tokens that align in lengths with their textual counterparts. However, the dependence on an external ASR system and the use of a non-causal decoder limits streaming use. To address this limitation, we propose TASTE-S, a streamable extension of TASTE suitable for real-time usage. TASTE-S integrates a CTC-based ASR module into the encoder for instant dual-modality encoding. We also redesign the unit decoder to enable on-the-fly decoding. With joint training, we show that TASTE-S matches TASTE's performance while significantly reducing latency. Further investigations reveal that TASTE-S remains robust to transcriptions and enables long-form encoding and decoding.
format Preprint
id arxiv_https___arxiv_org_abs_2603_12350
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TASTE-Streaming: Towards Streamable Text-Aligned Speech Tokenization and Embedding for Spoken Language Modeling
Tseng, Liang-Hsuan
Lee, Hung-yi
Computation and Language
Sound
Text-speech joint spoken language modeling (SLM) aims at natural and intelligent speech-based interactions, but developing such a system may suffer from modality mismatch: speech unit sequences are much longer than text tokens. Prior work reduces this gap with text-aligned tokenization and embedding (TASTE), producing speech tokens that align in lengths with their textual counterparts. However, the dependence on an external ASR system and the use of a non-causal decoder limits streaming use. To address this limitation, we propose TASTE-S, a streamable extension of TASTE suitable for real-time usage. TASTE-S integrates a CTC-based ASR module into the encoder for instant dual-modality encoding. We also redesign the unit decoder to enable on-the-fly decoding. With joint training, we show that TASTE-S matches TASTE's performance while significantly reducing latency. Further investigations reveal that TASTE-S remains robust to transcriptions and enables long-form encoding and decoding.
title TASTE-Streaming: Towards Streamable Text-Aligned Speech Tokenization and Embedding for Spoken Language Modeling
topic Computation and Language
Sound
url https://arxiv.org/abs/2603.12350