Guardado en:
Detalles Bibliográficos
Autores principales: Karani, Jash, Chittem, Adithya, Roy, Deepan, Joshi, Sandeep
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2602.22431
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914352127279104
author Karani, Jash
Chittem, Adithya
Roy, Deepan
Joshi, Sandeep
author_facet Karani, Jash
Chittem, Adithya
Roy, Deepan
Joshi, Sandeep
contents Millimeter-wave (mmWave) radar captures are band-limited and noisy, making for difficult reconstruction of intelligible full-bandwidth speech. In this work, we propose a two-stage speech reconstruction pipeline for mmWave using a Radar-Aware Dual-conditioned Generative Adversarial Network (RAD-GAN), which is capable of performing bandwidth extension on signals with low signal-to-noise ratios (-5 dB to -1 dB), captured through glass walls. We propose an mmWave-tailored Multi-Mel Discriminator (MMD) and a Residual Fusion Gate (RFG) to enhance the generator input to process multiple conditioning channels. The proposed two-stage pipeline involves pretraining the model on synthetically clipped clean speech and finetuning on fused mel spectrograms generated by the RFG. We empirically show that the proposed method, trained on a limited dataset, with no pre-trained modules, and no data augmentations, outperformed state-of-the-art approaches for this specific task. Audio examples of RAD-GAN are available online at https://rad-gan-demo-site.vercel.app/.
format Preprint
id arxiv_https___arxiv_org_abs_2602_22431
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle mmWave Radar Aware Dual-Conditioned GAN for Speech Reconstruction of Signals With Low SNR
Karani, Jash
Chittem, Adithya
Roy, Deepan
Joshi, Sandeep
Sound
Machine Learning
Millimeter-wave (mmWave) radar captures are band-limited and noisy, making for difficult reconstruction of intelligible full-bandwidth speech. In this work, we propose a two-stage speech reconstruction pipeline for mmWave using a Radar-Aware Dual-conditioned Generative Adversarial Network (RAD-GAN), which is capable of performing bandwidth extension on signals with low signal-to-noise ratios (-5 dB to -1 dB), captured through glass walls. We propose an mmWave-tailored Multi-Mel Discriminator (MMD) and a Residual Fusion Gate (RFG) to enhance the generator input to process multiple conditioning channels. The proposed two-stage pipeline involves pretraining the model on synthetically clipped clean speech and finetuning on fused mel spectrograms generated by the RFG. We empirically show that the proposed method, trained on a limited dataset, with no pre-trained modules, and no data augmentations, outperformed state-of-the-art approaches for this specific task. Audio examples of RAD-GAN are available online at https://rad-gan-demo-site.vercel.app/.
title mmWave Radar Aware Dual-Conditioned GAN for Speech Reconstruction of Signals With Low SNR
topic Sound
Machine Learning
url https://arxiv.org/abs/2602.22431