Masked Autoregressive Speech Enhancement with Continuous Neural Audio Codec Representations

• URL: http://arxiv.org/abs/2609.03940v1
• Date: Thu, 03 Sep 2026 14:45:42 GMT
• Title: Masked Autoregressive Speech Enhancement with Continuous Neural Audio Codec Representations
• Abstract summary: Masked autoregressive SE (MARSE) is a method for speech enhancement based on iterative decoding of clean speech frames.<n>Results show that MARSE enables a flexible trade-off between SE performance and computational cost.
• Score: 9.994255134903282
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Most previous work on speech enhancement (SE) based on masked generative modeling relied on discrete token representations of audio signals, obtained using neural audio codecs (NACs). However, a recent study has shown that continuous latent representations of NACs can be advantageous for SE in terms of speech quality and intelligibility. In this work, we propose masked autoregressive SE (MARSE), a method for SE based on iterative decoding of masked clean speech frames using continuous NAC representations of speech. In particular, we investigate a set of different decoding policies, ceteris paribus, that is, using the same DNN (a Conformer model), the same NAC (the DAC codec) and the same training setup. The results show that MARSE enables a flexible trade-off between SE performance and computational cost. Audio examples and code are available online.

Related papers

• EntangleCodec: A Unified Discrete Audio Tokenizer via Semantic-Acoustic Entanglement [54.95940885045993]
  EntangleCodec is a unified discrete audio tokenizer that learns caption-aligned semantic-acoustic representations before quantization.<n>It captures linguistic content, speaker identity, emotion, prosody, and acoustic scenes within a compact token stream.<n>It achieves reconstruction quality competitive with specialized codecs, and supports both TTS and TTA generation in a unified framework.
  arXiv  Detail & Related papers  (2026-06-01T18:05:18Z)
• AudioMosaic: Contrastive Masked Audio Representation Learning [53.52371029884106]
  We introduce textbfAudioMosaic, a contrastive learning-based audio encoder for general audio understanding.<n>AudioMosaic constructs positive pairs by applying structured time-frequency masking to spectrogram patches.<n>Experiments show that AudioMosaic achieves state-of-the-art performance on several standard audio benchmarks.
  arXiv  Detail & Related papers  (2026-05-14T00:56:51Z)
• Frontend Token Enhancement for Token-Based Speech Recognition [50.35062963870211]
  Discretized representations of speech signals are efficient alternatives to continuous features for speech recognition applications.<n>In this work, we introduce a system that estimates clean speech tokens from noisy speech and evaluate it on an ASR backend using semantic tokens.<n>We consider four types of enhancement models based on their input/token domains: wave-to-wave, token-to-output, continuous SSL features-to-token, and wave-to-token.
  arXiv  Detail & Related papers  (2026-02-04T05:02:15Z)
• UniSE: A Unified Framework for Decoder-only Autoregressive LM-based Speech Enhancement [3.855026553620411]
  We propose UniSE, a unified decoder-only LM-based framework to handle different speech enhancement tasks.<n>It takes input speech features as conditions and generates discrete tokens of the target speech using AR modeling.<n>Experiments indicate the proposed UniSE can achieve competitive performance compared to discriminative and generative baselines.
  arXiv  Detail & Related papers  (2025-10-23T11:22:24Z)
• FocalCodec: Low-Bitrate Speech Coding via Focal Modulation Networks [33.022035588157614]
  FocalCodec is an efficient low-bitrate based on focal modulation that utilizes a single binary codebook to compress speech.<n>Demo samples and code are available at https://lucadellalib.io/kbpscodec-web/.
  arXiv  Detail & Related papers  (2025-02-06T19:24:50Z)
• Codec-ASR: Training Performant Automatic Speech Recognition Systems with Discrete Speech Representations [16.577870835480585]
  We present a comprehensive analysis on building ASR systems with discrete codes. We investigate different methods for training such as quantization schemes and time-domain vs spectral feature encodings. We introduce a pipeline that outperforms Encodec at similar bit-rate.
  arXiv  Detail & Related papers  (2024-07-03T20:51:41Z)
• EnCodecMAE: Leveraging neural codecs for universal audio representation learning [16.590638305972632]
  We propose masking representations of the audio signal, and training a MAE to reconstruct the masked segments. We evaluate this approach, which we call EnCodecMAE, on a wide range of tasks involving speech, music and environmental sounds.
  arXiv  Detail & Related papers  (2023-09-14T02:21:53Z)
• TokenSplit: Using Discrete Speech Representations for Direct, Refined, and Transcript-Conditioned Speech Separation and Recognition [51.565319173790314]
  TokenSplit is a sequence-to-sequence encoder-decoder model that uses the Transformer architecture. We show that our model achieves excellent performance in terms of separation, both with or without transcript conditioning. We also measure the automatic speech recognition (ASR) performance and provide audio samples of speech synthesis to demonstrate the additional utility of our model.
  arXiv  Detail & Related papers  (2023-08-21T01:52:01Z)
• LA-VocE: Low-SNR Audio-visual Speech Enhancement using Neural Vocoders [53.30016986953206]
  We propose LA-VocE, a new two-stage approach that predicts mel-spectrograms from noisy audio-visual speech via a transformer-based architecture. We train and evaluate our framework on thousands of speakers and 11+ different languages, and study our model's ability to adapt to different levels of background noise and speech interference.
  arXiv  Detail & Related papers  (2022-11-20T15:27:55Z)
• Self-Supervised Learning for Speech Enhancement through Synthesis [5.924928860260821]
  We propose a denoising vocoder (DeVo) approach, where a vocoder accepts noisy representations and learns to directly synthesize clean speech. We demonstrate a causal version capable of running on streaming audio with 10ms latency and minimal performance degradation.
  arXiv  Detail & Related papers  (2022-11-04T16:06:56Z)
• Fast End-to-End Speech Recognition via a Non-Autoregressive Model and Cross-Modal Knowledge Transferring from BERT [72.93855288283059]
  We propose a non-autoregressive speech recognition model called LASO (Listen Attentively, and Spell Once) The model consists of an encoder, a decoder, and a position dependent summarizer (PDS)
  arXiv  Detail & Related papers  (2021-02-15T15:18:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.