paper-with-me

홈 › Papers

Attention-Constrained Inference for Robust Decoder-Only Text-to-Speech

2024-04-30 · Hankun Wang, Chenpeng Du, Yiwei Guo, Shuai Wang, Xie Chen, Kai Yu

Recent popular decoder-only text-to-speech models are known for their ability of generating natural-sounding speech. However, such models sometimes suffer from word skipping and repeating due to the lack of explicit monotonic alignment constraints. In this paper, we notice from the attention maps that some particular attention heads of the decoder-only model indicate the alignments between speech and text. We call the attention maps of those heads Alignment-Emerged Attention Maps (AEAMs). Based on this discovery, we propose a novel inference method without altering the training process, named Attention-Constrained Inference (ACI), to facilitate monotonic synthesis. It first identifies AEAMs using the Attention Sweeping algorithm and then applies constraining masks on AEAMs. Our experimental results on decoder-only TTS model VALL-E show that the WER of synthesized speech is reduced by up to 20.5% relatively with ACI while the naturalness and speaker similarity are comparable.

📄 PDF Abstract BibTeX arXiv:2404.19723

Code (0)

등록된 구현이 없습니다.

Tasks

Decodertext-to-speechText to Speech

Similar Papers 제목 키워드 기반

You Only Index Once: Cross-Layer Sparse Attention with Shared Routing

2026-06-04 · Yutao Sun, Yanqi Zhang, Li Dong, Jianyong Wang 외 arxiv

Long-context inference in modern LLMs is increasingly constrained by decoding efficiency, especially in reasoning-heavy settings where models generate long intermediate chains of thought. Existing sparse attention method…

Large Language Model Partitioning for Low-Latency Inference at the Edge

2025-05-05 · Dimitrios Kafetzis, Ramin Khalili, Iordanis Koutsopoulos

Large Language Models (LLMs) based on autoregressive, decoder-only Transformers generate text one token at a time, where a token represents a discrete unit of text. As each newly produced token is appended to the partial…

DecoderLanguage ModelingLanguage ModellingLarge Language Model

Parallel Refinements for Lexically Constrained Text Generation with BART

2021-09-26 · EMNLP 2021 11 · Xingwei He

Lexically constrained text generation aims to control the generated text by incorporating some pre-specified keywords into the output. Previous work injects lexical constraints into the output by controlling the decoding…

DecoderDiversitySentenceText Generation

You Only Cache Once: Decoder-Decoder Architectures for Language Models

2024-05-08 · Yutao Sun, Li Dong, Yi Zhu, Shaohan Huang 외

We introduce a decoder-decoder architecture, YOCO, for large language models, which only caches key-value pairs once. It consists of two components, i.e., a cross-decoder stacked upon a self-decoder. The self-decoder eff…

DecoderGPURetrieval

Block-Based Double Decoders

2026-05-11 · Asher Labovich, Benjamin Bradley, Vanessa Alexander, Chaitanya Harsha arxiv

Encoder-decoder models offer substantial inference-time savings over decoder-only models, but their pretraining objectives suffer from sparse supervision and dynamic sequence lengths, keeping them out of practice at scal…