Pith. sign in

REVIEW 5 cited by

TF-GridNet: Making Time-Frequency Domain Models Great Again for Monaural Speaker Separation

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2209.03952 v2 pith:ET2LXTAP submitted 2022-09-08 cs.SD eess.AS

Zhong-Qiu Wang , Samuele Cornell , Shukjae Choi , Younglo Lee , Byeong-Yeol Kim , Shinji Watanabe This is my paper · ORCID
classification cs.SDeess.AS
keywords inputmodelmoduleseparationcomponentsdomainlossmixture
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

We propose TF-GridNet, a novel multi-path deep neural network (DNN) operating in the time-frequency (T-F) domain, for monaural talker-independent speaker separation in anechoic conditions. The model stacks several multi-path blocks, each consisting of an intra-frame spectral module, a sub-band temporal module, and a full-band self-attention module, to leverage local and global spectro-temporal information for separation. The model is trained to perform complex spectral mapping, where the real and imaginary (RI) components of the input mixture are stacked as input features to predict target RI components. Besides using the scale-invariant signal-to-distortion ratio (SI-SDR) loss for model training, we include a novel loss term to encourage separated sources to add up to the input mixture. Without using dynamic mixing, we obtain 23.4 dB SI-SDR improvement (SI-SDRi) on the WSJ0-2mix dataset, outperforming the previous best by a large margin.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Ranking the Impact of Contextual Specialization in Neural Speech Enhancement

    eess.AS 2026-07 accept novelty 6.0 of 10

    Speaker-identity specialization produces the largest gains in neural speech enhancement, letting small models match or exceed generalists ten times their size.

  2. Making Separation-First Multi-Stream Audio Watermarking Feasible via Joint Training

    cs.SD 2026-03 conditional novelty 6.0 of 10

    Jointly training the watermark embedder/detector with the source separator enables ~1% bit-error-rate recovery of per-stem watermarks after mixing and separation, where independent training yields 15–35%.

  3. Speech Enhancement with Dual-path Multi-Channel Linear Prediction Filter and Multi-norm Beamforming

    eess.AS 2025-07 conditional novelty 6.0 of 10

    A dual-path time-frequency MCLP filter cascade plus a multi-norm (l2 and l1) beamformer improves speech enhancement metrics over classical baselines in simulated reverberant rooms, especially at high T60.

  4. TF-MossFormer: Integrating Convolution Gated Local-Global Attentions for Enhanced Time-Frequency Domain Monaural Speech Separation

    cs.SD 2026-07 conditional novelty 4.0 of 10

    A spectrogram-domain transformer combining fixed sliding-window local attention and global attention with convolutional gating reaches 24.4 dB SI-SDRi on WSJ0-2Mix at 25.4M parameters.

  5. Developing an Effective Training Dataset to Enhance the Performance of AI-based Speaker Separation Systems

    cs.SD 2024-11 conditional novelty 4.0 of 10

    A playback-and-record method creates a realistic two-speaker training set that yields up to 1.65 dB SI-SDR improvement over synthetic training.

Pith tools