CTC-TTS: LLM-based dual-streaming text-to-speech with CTC alignment

TL;DR

This paper introduces CTC-TTS, a novel dual-streaming text-to-speech system using CTC alignment and bi-word interleaving, achieving better quality and lower latency than previous methods.

Contribution

It replaces traditional MFA alignment with a CTC-based aligner and proposes two variants for different quality-latency trade-offs, advancing low-latency neural TTS.

Findings

Outperforms MFA-based and fixed-ratio methods in streaming synthesis

Achieves higher quality in zero-shot TTS tasks

Demonstrates effective low-latency synthesis with CTC-TTS-F

Abstract

Large-language-model (LLM)-based text-to-speech (TTS) systems can generate natural speech, but most are not designed for low-latency dual-streaming synthesis. High-quality dual-streaming TTS depends on accurate text--speech alignment and well-designed training sequences that balance synthesis quality and latency. Prior work often relies on GMM-HMM based forced-alignment toolkits (e.g., MFA), which are pipeline-heavy and less flexible than neural aligners; fixed-ratio interleaving of text and speech tokens struggles to capture text--speech alignment regularities. We propose CTC-TTS, which replaces MFA with a CTC based aligner and introduces a bi-word based interleaving strategy. Two variants are designed: CTC-TTS-L (token concatenation along the sequence length) for higher quality and CTC-TTS-F (embedding stacking along the feature dimension) for lower latency. Experiments show that CTC-TTS outperforms fixed-ratio interleaving and MFA-based baselines on streaming synthesis and zero-shot tasks. Speech samples are available at https://ctctts.github.io/.