Accelerating Codec-based Speech Synthesis With Multi-token Prediction And Speculative Decoding

Abstract

The goal of this paper is to accelerate codec-based speech synthesis systems with minimum sacrifice to speech quality. We propose an enhanced inference method that allows for flexible trade-offs between speed and quality during inference without requiring additional training. Our core idea is to predict multiple tokens per inference step of the AR module using multiple prediction heads, resulting in a linear reduction in synthesis time as the number of heads increases. Furthermore, we introduce a novel speculative decoding technique that utilises a Viterbi-based algorithm to select the optimal sequence of generated tokens at each decoding step. In our experiments, we demonstrate that the time required to predict each token is reduced by a factor of 4 to 5 compared to baseline models, with minimal quality trade-off or even improvement in terms of speech intelligibility. Audio samples are available at: multpletokensprediction.github.io/multipletokensprediction.github.io/.

Stats

• citations3
• S2 citations—
• github stars0
• HF likes0
• heat score4.52
• arxiv keynguyen2024accelerating

Related papers

• Fast And High-quality Auto-regressive Speech Synthesis Via Speculative Decoding (2024)5.24
• Spectral Codecs: Improving Non-autoregressive Speech Synthesis With Spectrogram-based Audio Codecs (2024)0.00
• Principled Coarse-grained Acceptance For Speculative Decoding In Speech (2025)0.00
• Livespeech: Low-latency Zero-shot Text-to-speech Via Autoregressive Modeling Of Audio Discrete Codes (2024)5.84
• Single-codec: Single-codebook Speech Codec Towards High-performance Speech Generation (2024)9.23
• Msr-codec: A Low-bitrate Multi-stream Residual Codec For High-fidelity Speech Generation With Information Disentanglement (2025)2.35
• Socodec: A Semantic-ordered Multi-stream Speech Codec For Efficient Language Model Based Text-to-speech Synthesis (2024)6.34
• Investigating Neural Audio Codecs For Speech Language Model-based Speech Generation (2024)2.26