In this work, we propose a novel consistency-preserving loss function for recovering the phase information in the context of phase reconstruction (PR) and speech enhancement (SE). Different from conventional techniques that directly estimate the phase using a deep model, our idea is to exploit ad-hoc constraints to directly generate a consistent pair of magnitude and phase. Specifically, the proposed loss forces a set of complex numbers to be a consistent short-time Fourier transform (STFT) representation, i.e., to be the spectrogram of a real signal. Our approach thus avoids the difficulty of estimating the original phase, which is highly unstructured and sensitive to time shift. The influence of our proposed loss is first assessed on a PR task, experimentally demonstrating that our approach is viable. Next, we show its effectiveness on an SE task, using both the VB-DMD and WSJ0-CHiME3 data sets. On VB-DMD, our approach is competitive with conventional solutions. On the challenging WSJ0-CHiME3 set, the proposed framework compares favourably over those techniques that explicitly estimate the phase.

This paper proposes a generative pretraining foundation model for high-quality speech restoration tasks. By directly operating on complex-valued short-time Fourier transform coefficients, our model does not rely on any vocoders for time-domain signal reconstruction. As a result, our model simplifies the synthesis process and removes the quality upper-bound introduced by any mel-spectrogram vocoder compared to prior work SpeechFlow. The proposed method is evaluated on multiple speech restoration tasks, including speech denoising, bandwidth extension, codec artifact removal, and target speaker extraction. In all scenarios, finetuning our pretrained model results in superior performance over strong baselines. Notably, in the target speaker extraction task, our model outperforms existing systems, including those leveraging SSL-pretrained encoders like WavLM. The code and the pretrained checkpoints are publicly available in the NVIDIA NeMo framework.

This paper proposes two novel variants of neural reprogramming to enhance wake word recognition in streaming end-to-end ASR models without updating model weights. The first, "trigger-frame reprogramming", prepends the input speech feature sequence with the learned trigger-frames of the target wake word to adjust ASR model’s hidden states for improved wake word recognition. The second, "predictor-state initialization", trains only the initial state vectors (cell and hidden states) of the LSTMs in the prediction network. When applying to a baseline LibriSpeech Emformer RNN-T model with a 98% wake word verification false rejection rate (FRR) on unseen wake words, the proposed approaches achieve 76% and 97% relative FRR reductions with no increase on false acceptance rate. In-depth characteristic analyses of the proposed approaches are also conducted to provide deeper insights. These approaches offer an effective hot-fixing methods to improve wake word recognition performance in deployed production ASR models without the need for model updates.

We propose a novel language-universal approach to end-to-end automatic spoken keyword recognition (SKR) leveraging upon (i) a self-supervised pre-trained model, and (ii) a set of universal speech attributes (manner and place of articulation).Specifically, Wav2Vec2.0 is used to generate robust speech representations, followed by a linear output layer to produce attribute sequences.A non-trainable pronunciation model then maps sequences of attributes into spoken keywords in a multilingual setting.Experiments on the Multilingual Spoken Words Corpus show comparable performances to character-and phoneme-based SKR in seen languages.The inclusion of domain adversarial training (DAT) improves the proposed framework, outperforming both character-and phoneme-based SKR approaches with 13.73% and 17.22% relative word error rate (WER) reduction in seen languages, and achieves 32.14% and 19.92% WER reduction for unseen languages in zero-shot settings.

We propose a multi-dimensional structured state space (S4) approach to speech enhancement. To better capture the spectral dependencies across the frequency axis, we focus on modifying the multi-dimensional S4 layer with whitening transformation to build new small-footprint models that also achieve good performance. We explore several S4-based deep architectures in time (T) and time-frequency (TF) domains. The 2-D S4 layer can be considered a particular convolutional layer with an infinite receptive field although it utilizes fewer parameters than a conventional convolutional layer. Evaluated on the VoiceBank-DEMAND data set, when compared with the conventional U-net model based on convolutional layers, the proposed TF-domain S4-based model is 78.6% smaller in size, yet it still achieves competitive results with a PESQ score of 3.15 with data augmentation. By increasing the model size, we can even reach a PESQ score of 3.18.

We propose a novel adversarial reprogramming (AR) approach for low-resource spoken command recognition (SCR) and build an AR-SCR system. The AR procedure aims at repurposing a pretrained SCR model (from the source domain) to modify the acoustic signals (from the target domain). To solve the label mismatches between source and target domains and further improve the stability of AR, we propose a novel similarity-based label mapping technique to align classes. In addition, the transfer learning (TL) technique is combined with the original AR process to improve the model adaptation capability. We evaluate the proposed AR-SCR system on three low-resource SCR datasets, including Arabic, Lithuanian, and dysarthric Mandarin speech. Experimental results show that with a pretrained acoustic model trained on a large-scale English dataset, the proposed AR-SCR system outperforms the current state-of-the-art results on Lithuanian and Arabic datasets, with only a limited amount of training data.