Electrical Engineering And Systems Science

In this paper, we formulate a blind source separation (BSS) framework, which allows integrating U-Net based deep learning source separation network with probabilistic spatial machine learning expectation maximization (EM) algorithm for separating speech in reverberant conditions. Our proposed model uses a pre-trained deep learning convolutional neural network, U-Net, for clustering the interaural level difference (ILD) cues and machine learning expectation maximization (EM) algorithm for clustering the interaural phase difference (IPD) cues. The integrated model exploits the complementary strengths of the two approaches to BSS: the strong modeling power of supervised neural networks and the ease of unsupervised machine learning algorithms, whose few parameters can be estimated on as little as a single segment of an audio mixture. The results show an average improvement of 4.3 dB in signal to distortion ratio (SDR) and 4.3% in short time speech intelligibility (STOI) over the EM based source separation algorithm MESSL-GS (model-based expectation-maximization source separation and localization with garbage source) and 4.5 dB in SDR and 8% in STOI over deep learning convolutional neural network (U-Net) based speech separation algorithm SONET under the reverberant conditions ranging from anechoic to those mostly encountered in the real world.

We present a simple and efficient auxiliary loss function for automatic speech recognition (ASR) based on the connectionist temporal classification (CTC) objective. The proposed objective, an intermediate CTC loss, is attached to an intermediate layer in the CTC encoder network. This intermediate CTC loss well regularizes CTC training and improves the performance requiring only small modification of the code and small and no overhead during training and inference, respectively. In addition, we propose to combine this intermediate CTC loss with stochastic depth training, and apply this combination to a recently proposed Conformer network. We evaluate the proposed method on various corpora, reaching word error rate (WER) 9.9% on the WSJ corpus and character error rate (CER) 5.2% on the AISHELL-1 corpus respectively, based on CTC greedy search without a language model. Especially, the AISHELL-1 task is comparable to other state-of-the-art ASR systems based on auto-regressive decoder with beam search.

The efficacy of external language model (LM) integration with existing end-to-end (E2E) automatic speech recognition (ASR) systems can be improved significantly using the internal language model estimation (ILME) method. In this method, the internal LM score is subtracted from the score obtained by interpolating the E2E score with the external LM score, during inference. To improve the ILME-based inference, we propose an internal LM training (ILMT) method to minimize an additional internal LM loss by updating only the E2E model components that affect the internal LM estimation. ILMT encourages the E2E model to form a standalone LM inside its existing components, without sacrificing ASR accuracy. After ILMT, the more modular E2E model with matched training and inference criteria enables a more thorough elimination of the source-domain internal LM, and therefore leads to a more effective integration of the target-domain external LM. Experimented with 30K-hour trained recurrent neural network transducer and attention-based encoder-decoder models, ILMT with ILME-based inference achieves up to 31.5% and 11.4% relative word error rate reductions from standard E2E training with Shallow Fusion on out-of-domain LibriSpeech and in-domain Microsoft production test sets, respectively.

Knowledge Distillation (KD) is a popular area of research for reducing the size of large models while still maintaining good performance. The outputs of larger teacher models are used to guide the training of smaller student models. Given the repetitive nature of acoustic events, we propose to leverage this information to regulate the KD training for Audio Tagging. This novel KD method, "Intra-Utterance Similarity Preserving KD" (IUSP), shows promising results for the audio tagging task. It is motivated by the previously published KD method: "Similarity Preserving KD" (SP). However, instead of preserving the pairwise similarities between inputs within a mini-batch, our method preserves the pairwise similarities between the frames of a single input utterance. Our proposed KD method, IUSP, shows consistent improvements over SP across student models of different sizes on the DCASE 2019 Task 5 dataset for audio tagging. There is a 27.1% to 122.4% percent increase in improvement of micro AUPRC over the baseline relative to SP's improvement of over the baseline.

In this paper, we propose an effective training strategy to ex-tract robust speaker representations from a speech signal. Oneof the key challenges in speaker recognition tasks is to learnlatent representations or embeddings containing solely speakercharacteristic information in order to be robust in terms of intra-speaker variations. By modifying the network architecture togenerate both speaker-related and speaker-unrelated representa-tions, we exploit a learning criterion which minimizes the mu-tual information between these disentangled embeddings. Wealso introduce an identity change loss criterion which utilizes areconstruction error to different utterances spoken by the samespeaker. Since the proposed criteria reduce the variation ofspeaker characteristics caused by changes in background envi-ronment or spoken content, the resulting embeddings of eachspeaker become more consistent. The effectiveness of the pro-posed method is demonstrated through two tasks; disentangle-ment performance, and improvement of speaker recognition ac-curacy compared to the baseline model on a benchmark dataset,VoxCeleb1. Ablation studies also show the impact of each cri-terion on overall performance.

This study investigates the speech articulatory coordination in schizophrenia subjects exhibiting strong positive symptoms (e.g.hallucinations and delusions), using a time delay embedded correlation analysis. We show that the schizophrenia subjects with strong positive symptoms and who are markedly ill pose complex coordination patterns in facial and speech gestures than what is observed in healthy subjects. This observation is in contrast to what previous studies have shown in Major Depressive Disorder (MDD), where subjects with MDD show a simpler coordination pattern with respect to healthy controls or subjects in remission. This difference is not surprising given MDD is necessarily accompanied by Psychomotor slowing (i.e.,negative symptoms) which affects speech, ideation and motility. With respect to speech, psychomotor slowing results in slowed speech with more and longer pauses than what occurs in speech from the same speaker when they are in remission and from a healthy subject. Time delay embedded correlation analysis has been used to quantify the differences in coordination patterns of speech articulation. The current study is based on 17 Facial Action Units (FAUs) extracted from video data and 6 Vocal Tract Variables (TVs) obtained from simultaneously recorded audio data. The TVs are extracted using a speech inversion system based on articulatory phonology that maps the acoustic signal to vocal tract variables. The high-level time delay embedded correlation features computed from TVs and FAUs are used to train a stacking ensemble classifier fusing audio and video modalities. The results show that there is a promising distinction between healthy and schizophrenia subjects (with strong positive symptoms) in terms of neuromotor coordination in speech.

Recently, an end-to-end (E2E) speaker-attributed automatic speech recognition (SA-ASR) model was proposed as a joint model of speaker counting, speech recognition and speaker identification for monaural overlapped speech. It showed promising results for simulated speech mixtures consisting of various numbers of speakers. However, the model required prior knowledge of speaker profiles to perform speaker identification, which significantly limited the application of the model. In this paper, we extend the prior work by addressing the case where no speaker profile is available. Specifically, we perform speaker counting and clustering by using the internal speaker representations of the E2E SA-ASR model to diarize the utterances of the speakers whose profiles are missing from the speaker inventory. We also propose a simple modification to the reference labels of the E2E SA-ASR training which helps handle continuous multi-talker recordings well. We conduct a comprehensive investigation of the original E2E SA-ASR and the proposed method on the monaural LibriCSS dataset. Compared to the original E2E SA-ASR with relevant speaker profiles, the proposed method achieves a close performance without any prior speaker knowledge. We also show that the source-target attention in the E2E SA-ASR model provides information about the start and end times of the hypotheses.

Phase serves as a critical component of speech that influences the quality and intelligibility. Current speech enhancement algorithms are beginning to address phase distortions, but the algorithms focus on normal-hearing (NH) listeners. It is not clear whether phase enhancement is beneficial for hearing-impaired (HI) listeners. We investigated the influence of phase distortion on speech quality through a listening study, in which NH and HI listeners provided speech-quality ratings using the MUSHRA procedure. In one set of conditions, the speech was mixed with babble noise at 4 different signal-to-noise ratios (SNRs) from -5 to 10 dB. In another set of conditions, the SNR was fixed at 10 dB and the noisy speech was presented in a simulated reverberant room with T60s ranging from 100 to 1000 ms. The speech level was kept at 65 dB SPL for NH listeners and amplification was applied for HI listeners to ensure audibility. Ideal ratio masking (IRM) was used to simulate speech enhancement. Two objective metrics (i.e., PESQ and HASQI) were utilized to compare subjective and objective ratings. Results indicate that phase distortion has a negative impact on perceived quality for both groups and PESQ is more closely correlated with human ratings.

End-to-end models are fast replacing conventional hybrid models in automatic speech recognition. A transformer is a sequence-to-sequence framework solely based on attention, that was initially applied to machine translation task. This end-to-end framework has been shown to give promising results when used for automatic speech recognition as well. In this paper, we explore different ways of incorporating speaker information while training a transformer-based model to improve its performance. We present speaker information in the form of speaker embeddings for each of the speakers. Two broad categories of speaker embeddings are used: (i)fixed embeddings, and (ii)learned embeddings. We experiment using speaker embeddings learned along with the model training, as well as one-hot vectors and x-vectors. Using these different speaker embeddings, we obtain an average relative improvement of 1% to 3% in the token error rate. We report results on the NPTEL lecture database. NPTEL is an open-source e-learning portal providing content from top Indian universities.

We propose a new algorithm for joint dereverberation and blind source separation (DR-BSS). Our work builds upon the IRLMA-T framework that applies a unified filter combining dereverberation and separation. One drawback of this framework is that it requires several matrix inversions, an operation inherently costly and with potential stability issues. We leverage the recently introduced iterative source steering (ISS) updates to propose two algorithms mitigating this issue. Albeit derived from first principles, the first algorithm turns out to be a natural combination of weighted prediction error (WPE) dereverberation and ISS-based BSS, applied alternatingly. In this case, we manage to reduce the number of matrix inversion to only one per iteration and source. The second algorithm updates the ILRMA-T matrix using only sequential ISS updates requiring no matrix inversion at all. Its implementation is straightforward and memory efficient. Numerical experiments demonstrate that both methods achieve the same final performance as ILRMA-T in terms of several relevant objective metrics. In the important case of two sources, the number of iterations required is also similar.