We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Audio and Speech Processing

Title: Time-domain speech super-resolution with GAN based modeling for telephony speaker verification

Abstract: Automatic Speaker Verification (ASV) technology has become commonplace in virtual assistants. However, its performance suffers when there is a mismatch between the train and test domains. Mixed bandwidth training, i.e., pooling training data from both domains, is a preferred choice for developing a universal model that works for both narrowband and wideband domains. We propose complementing this technique by performing neural upsampling of narrowband signals, also known as bandwidth extension. Our main goal is to discover and analyze high-performing time-domain Generative Adversarial Network (GAN) based models to improve our downstream state-of-the-art ASV system. We choose GANs since they (1) are powerful for learning conditional distribution and (2) allow flexible plug-in usage as a pre-processor during the training of downstream task (ASV) with data augmentation. Prior works mainly focus on feature-domain bandwidth extension and limited experimental setups. We address these limitations by 1) using time-domain extension models, 2) reporting results on three real test sets, 2) extending training data, and 3) devising new test-time schemes. We compare supervised (conditional GAN) and unsupervised GANs (CycleGAN) and demonstrate average relative improvement in Equal Error Rate of 8.6% and 7.7%, respectively. For further analysis, we study changes in spectrogram visual quality, audio perceptual quality, t-SNE embeddings, and ASV score distributions. We show that our bandwidth extension leads to phenomena such as a shift of telephone (test) embeddings towards wideband (train) signals, a negative correlation of perceptual quality with downstream performance, and condition-independent score calibration.
Comments: Submit to IEEE/ACM Transactions on Audio, Speech, and Language Processing
Subjects: Audio and Speech Processing (eess.AS)
Cite as: arXiv:2209.01702 [eess.AS]
  (or arXiv:2209.01702v1 [eess.AS] for this version)

Submission history

From: Saurabh Kataria [view email]
[v1] Sun, 4 Sep 2022 22:34:41 GMT (5483kb,D)

Link back to: arXiv, form interface, contact.