Chao Zhang

Changli Tang, Wenyi Yu, Guangzhi Sun et al.

Hearing is arguably an essential ability of artificial intelligence (AI) agents in the physical world, which refers to the perception and understanding of general auditory information consisting of at least three types of sounds: speech, audio events, and music. In this paper, we propose SALMONN, a speech audio language music open neural network, built by integrating a pre-trained text-based large language model (LLM) with speech and audio encoders into a single multimodal model. SALMONN enables the LLM to directly process and understand general audio inputs and achieve competitive performances on a number of speech and audio tasks used in training, such as automatic speech recognition and translation, auditory-information-based question answering, emotion recognition, speaker verification, and music and audio captioning etc. SALMONN also has a diverse set of emergent abilities unseen in the training, which includes but is not limited to speech translation to untrained languages, speech-based slot filling, spoken-query-based question answering, audio-based storytelling, and speech audio co-reasoning etc. The presence of cross-modal emergent abilities is studied, and a novel few-shot activation tuning approach is proposed to activate such abilities. To our knowledge, SALMONN is the first model of its type and can be regarded as a step towards AI with generic hearing abilities. The source code, model checkpoints and data are available at https://github.com/bytedance/SALMONN.

Chao-Han Huck Yang, Taejin Park, Yuan Gong et al. · gatech

Given recent advances in generative AI technology, a key question is how large language models (LLMs) can enhance acoustic modeling tasks using text decoding results from a frozen, pretrained automatic speech recognition (ASR) model. To explore new capabilities in language modeling for speech processing, we introduce the generative speech transcription error correction (GenSEC) challenge. This challenge comprises three post-ASR language modeling tasks: (i) post-ASR transcription correction, (ii) speaker tagging, and (iii) emotion recognition. These tasks aim to emulate future LLM-based agents handling voice-based interfaces while remaining accessible to a broad audience by utilizing open pretrained language models or agent-based APIs. We also discuss insights from baseline evaluations, as well as lessons learned for designing future evaluations.

Siyin Wang, Chao-Han Huck Yang, Ji Wu et al. · nvidia

This paper investigates the in-context learning abilities of the Whisper automatic speech recognition (ASR) models released by OpenAI. A novel speech-based in-context learning (SICL) approach is proposed for test-time adaptation, which can reduce the word error rates (WERs) with only a small number of labelled speech samples without gradient descent. Language-level adaptation experiments using Chinese dialects showed that when applying SICL to isolated word ASR, consistent and considerable relative WER reductions can be achieved using Whisper models of any size on two dialects, which is on average 32.3%. A k-nearest-neighbours-based in-context example selection technique can be applied to further improve the efficiency of SICL, which can increase the average relative WER reduction to 36.4%. The findings are verified using speaker adaptation or continuous speech recognition tasks, and both achieved considerable relative WER reductions. Detailed quantitative analyses are also provided to shed light on SICL's adaptability to phonological variances and dialect-specific lexical nuances.

Wen Wu, Wenlin Chen, Chao Zhang et al. · cambridge

Human annotator simulation (HAS) serves as a cost-effective substitute for human evaluation such as data annotation and system assessment. Human perception and behaviour during human evaluation exhibit inherent variability due to diverse cognitive processes and subjective interpretations, which should be taken into account in modelling to better mimic the way people perceive and interact with the world. This paper introduces a novel meta-learning framework that treats HAS as a zero-shot density estimation problem, which incorporates human variability and allows for the efficient generation of human-like annotations for unlabelled test inputs. Under this framework, we propose two new model classes, conditional integer flows and conditional softmax flows, to account for ordinal and categorical annotations, respectively. The proposed method is evaluated on three real-world human evaluation tasks and shows superior capability and efficiency to predict the aggregated behaviours of human annotators, match the distribution of human annotations, and simulate the inter-annotator disagreements.

Guangzhi Sun, Xianrui Zheng, Chao Zhang et al.

End-to-end automatic speech recognition (ASR) and large language models, such as Whisper and GPT-2, have recently been scaled to use vast amounts of training data. Despite the large amount of training data, infrequent content words that occur in a particular task may still exhibit poor ASR performance, with contextual biasing a possible remedy. This paper investigates the effectiveness of neural contextual biasing for Whisper combined with GPT-2. Specifically, this paper proposes integrating an adapted tree-constrained pointer generator (TCPGen) component for Whisper and a dedicated training scheme to dynamically adjust the final output without modifying any Whisper model parameters. Experiments across three datasets show a considerable reduction in errors on biasing words with a biasing list of 1000 words. Contextual biasing was more effective when applied to domain-specific data and can boost the performance of Whisper and GPT-2 without losing their generality.

Shuo-yiin Chang, Bo Li, Tara N. Sainath et al.

While a streaming voice assistant system has been used in many applications, this system typically focuses on unnatural, one-shot interactions assuming input from a single voice query without hesitation or disfluency. However, a common conversational utterance often involves multiple queries with turn-taking, in addition to disfluencies. These disfluencies include pausing to think, hesitations, word lengthening, filled pauses and repeated phrases. This makes doing speech recognition with conversational speech, including one with multiple queries, a challenging task. To better model the conversational interaction, it is critical to discriminate disfluencies and end of query in order to allow the user to hold the floor for disfluencies while having the system respond as quickly as possible when the user has finished speaking. In this paper, we present a turntaking predictor built on top of the end-to-end (E2E) speech recognizer. Our best system is obtained by jointly optimizing for ASR task and detecting when the user is paused to think or finished speaking. The proposed approach demonstrates over 97% recall rate and 85% precision rate on predicting true turn-taking with only 100 ms latency on a test set designed with 4 types of disfluencies inserted in conversational utterances.

Guangzhi Sun, Chao Zhang, Philip C. Woodland

Incorporating biasing words obtained as contextual knowledge is critical for many automatic speech recognition (ASR) applications. This paper proposes the use of graph neural network (GNN) encodings in a tree-constrained pointer generator (TCPGen) component for end-to-end contextual ASR. By encoding the biasing words in the prefix-tree with a tree-based GNN, lookahead for future wordpieces in end-to-end ASR decoding is achieved at each tree node by incorporating information about all wordpieces on the tree branches rooted from it, which allows a more accurate prediction of the generation probability of the biasing words. Systems were evaluated on the Librispeech corpus using simulated biasing tasks, and on the AMI corpus by proposing a novel visual-grounded contextual ASR pipeline that extracts biasing words from slides alongside each meeting. Results showed that TCPGen with GNN encodings achieved about a further 15% relative WER reduction on the biasing words compared to the original TCPGen, with a negligible increase in the computation cost for decoding.

Wen Wu, Chao Zhang, Xixin Wu et al.

Emotion recognition is a key attribute for artificial intelligence systems that need to naturally interact with humans. However, the task definition is still an open problem due to the inherent ambiguity of emotions. In this paper, a novel Bayesian training loss based on per-utterance Dirichlet prior distributions is proposed for verbal emotion recognition, which models the uncertainty in one-hot labels created when human annotators assign the same utterance to different emotion classes. An additional metric is used to evaluate the performance by detection test utterances with high labelling uncertainty. This removes a major limitation that emotion classification systems only consider utterances with labels where the majority of annotators agree on the emotion class. Furthermore, a frequentist approach is studied to leverage the continuous-valued "soft" labels obtained by averaging the one-hot labels. We propose a two-branch model structure for emotion classification on a per-utterance basis, which achieves state-of-the-art classification results on the widely used IEMOCAP dataset. Based on this, uncertainty estimation experiments were performed. The best performance in terms of the area under the precision-recall curve when detecting utterances with high uncertainty was achieved by interpolating the Bayesian training loss with the Kullback-Leibler divergence training loss for the soft labels. The generality of the proposed approach was verified using the MSP-Podcast dataset which yielded the same pattern of results.

Wen Wu, Chao Zhang, Philip C. Woodland

In automatic emotion recognition (AER), labels assigned by different human annotators to the same utterance are often inconsistent due to the inherent complexity of emotion and the subjectivity of perception. Though deterministic labels generated by averaging or voting are often used as the ground truth, it ignores the intrinsic uncertainty revealed by the inconsistent labels. This paper proposes a Bayesian approach, deep evidential emotion regression (DEER), to estimate the uncertainty in emotion attributes. Treating the emotion attribute labels of an utterance as samples drawn from an unknown Gaussian distribution, DEER places an utterance-specific normal-inverse gamma prior over the Gaussian likelihood and predicts its hyper-parameters using a deep neural network model. It enables a joint estimation of emotion attributes along with the aleatoric and epistemic uncertainties. AER experiments on the widely used MSP-Podcast and IEMOCAP datasets showed DEER produced state-of-the-art results for both the mean values and the distribution of emotion attributes.

Guangzhi Sun, Chao Zhang, Philip C Woodland

Contextual knowledge is essential for reducing speech recognition errors on high-valued long-tail words. This paper proposes a novel tree-constrained pointer generator (TCPGen) component that enables end-to-end ASR models to bias towards a list of long-tail words obtained using external contextual information. With only a small overhead in memory use and computation cost, TCPGen can structure thousands of biasing words efficiently into a symbolic prefix-tree and creates a neural shortcut between the tree and the final ASR output to facilitate the recognition of the biasing words. To enhance TCPGen, we further propose a novel minimum biasing word error (MBWE) loss that directly optimises biasing word errors during training, along with a biasing-word-driven language model discounting (BLMD) method during the test. All contextual ASR systems were evaluated on the public Librispeech audiobook corpus and the data from the dialogue state tracking challenges (DSTC) with the biasing lists extracted from the dialogue-system ontology. Consistent word error rate (WER) reductions were achieved with TCPGen, which were particularly significant on the biasing words with around 40\% relative reductions in the recognition error rates. MBWE and BLMD further improved the effectiveness of TCPGen and achieved more significant WER reductions on the biasing words. TCPGen also achieved zero-shot learning of words not in the audio training set with large WER reductions on the out-of-vocabulary words in the biasing list.

Guangzhi Sun, Chao Zhang, Philip C. Woodland

End-to-end spoken language understanding (SLU) suffers from the long-tail word problem. This paper exploits contextual biasing, a technique to improve the speech recognition of rare words, in end-to-end SLU systems. Specifically, a tree-constrained pointer generator (TCPGen), a powerful and efficient biasing model component, is studied, which leverages a slot shortlist with corresponding entities to extract biasing lists. Meanwhile, to bias the SLU model output slot distribution, a slot probability biasing (SPB) mechanism is proposed to calculate a slot distribution from TCPGen. Experiments on the SLURP dataset showed consistent SLU-F1 improvements using TCPGen and SPB, especially on unseen entities. On a new split by holding out 5 slot types for the test, TCPGen with SPB achieved zero-shot learning with an SLU-F1 score over 50% compared to baselines which can not deal with it. In addition to slot filling, the intent classification accuracy was also improved.

Ziyang Zhang, Andrew Thwaites, Alexandra Woolgar et al.

In complex auditory environments, the human auditory system possesses the remarkable ability to focus on a specific speaker while disregarding others. In this study, a new model named SWIM, a short-window convolution neural network (CNN) integrated with Mamba, is proposed for identifying the locus of auditory attention (left or right) from electroencephalography (EEG) signals without relying on speech envelopes. SWIM consists of two parts. The first is a short-window CNN (SW$_\text{CNN}$), which acts as a short-term EEG feature extractor and achieves a final accuracy of 84.9% in the leave-one-speaker-out setup on the widely used KUL dataset. This improvement is due to the use of an improved CNN structure, data augmentation, multitask training, and model combination. The second part, Mamba, is a sequence model first applied to auditory spatial attention decoding to leverage the long-term dependency from previous SW$_\text{CNN}$ time steps. By joint training SW$_\text{CNN}$ and Mamba, the proposed SWIM structure uses both short-term and long-term information and achieves an accuracy of 86.2%, which reduces the classification errors by a relative 31.0% compared to the previous state-of-the-art result. The source code is available at https://github.com/windowso/SWIM-ASAD.

Zhimin Li, Jianwei Zhang, Qin Lin et al.

We present Hunyuan-DiT, a text-to-image diffusion transformer with fine-grained understanding of both English and Chinese. To construct Hunyuan-DiT, we carefully design the transformer structure, text encoder, and positional encoding. We also build from scratch a whole data pipeline to update and evaluate data for iterative model optimization. For fine-grained language understanding, we train a Multimodal Large Language Model to refine the captions of the images. Finally, Hunyuan-DiT can perform multi-turn multimodal dialogue with users, generating and refining images according to the context. Through our holistic human evaluation protocol with more than 50 professional human evaluators, Hunyuan-DiT sets a new state-of-the-art in Chinese-to-image generation compared with other open-source models. Code and pretrained models are publicly available at github.com/Tencent/HunyuanDiT

Ziyun Cui, Wen Wu, Wei-Qiang Zhang et al.

The detection of Alzheimer's disease (AD) from spontaneous speech has attracted increasing attention while the sparsity of training data remains an important issue. This paper handles the issue by knowledge transfer, specifically from both speech-generic and depression-specific knowledge. The paper first studies sequential knowledge transfer from generic foundation models pretrained on large amounts of speech and text data. A block-wise analysis is performed for AD diagnosis based on the representations extracted from different intermediate blocks of different foundation models. Apart from the knowledge from speech-generic representations, this paper also proposes to simultaneously transfer the knowledge from a speech depression detection task based on the high comorbidity rates of depression and AD. A parallel knowledge transfer framework is studied that jointly learns the information shared between these two tasks. Experimental results show that the proposed method improves AD and depression detection, and produces a state-of-the-art F1 score of 0.928 for AD diagnosis on the commonly used ADReSSo dataset.

Yuchen Hu, Chen Chen, Chao-Han Huck Yang et al.

We propose an unsupervised adaptation framework, Self-TAught Recognizer (STAR), which leverages unlabeled data to enhance the robustness of automatic speech recognition (ASR) systems in diverse target domains, such as noise and accents. STAR is developed for prevalent speech foundation models based on Transformer-related architecture with auto-regressive decoding (e.g., Whisper, Canary). Specifically, we propose a novel indicator that empirically integrates step-wise information during decoding to assess the token-level quality of pseudo labels without ground truth, thereby guiding model updates for effective unsupervised adaptation. Experimental results show that STAR achieves an average of 13.5% relative reduction in word error rate across 14 target domains, and it sometimes even approaches the upper-bound performance of supervised adaptation. Surprisingly, we also observe that STAR prevents the adapted model from the common catastrophic forgetting problem without recalling source-domain data. Furthermore, STAR exhibits high data efficiency that only requires less than one-hour unlabeled data, and seamless generality to alternative large speech models and speech translation tasks. Our code aims to open source to the research communities.

Qinfan Xiao, Ziyun Cui, Chi Zhang et al.

Electroencephalography (EEG) and magnetoencephalography (MEG) measure neural activity non-invasively by capturing electromagnetic fields generated by dendritic currents. Although rooted in the same biophysics, EEG and MEG exhibit distinct signal patterns, further complicated by variations in sensor configurations across modalities and recording devices. Existing approaches typically rely on separate, modality- and dataset-specific models, which limits the performance and cross-domain scalability. This paper proposes BrainOmni, the first brain foundation model that generalises across heterogeneous EEG and MEG recordings. To unify diverse data sources, we introduce BrainTokenizer,the first tokenizer that quantises spatiotemporal brain activity into discrete representations. Central to BrainTokenizer is a novel Sensor Encoder that encodes sensor properties such as spatial layout, orientation, and type, enabling compatibility across devices and modalities. Building upon the discrete representations, BrainOmni learns unified semantic embeddings of brain signals by self-supervised pretraining. To the best of our knowledge, it is the first foundation model to support both EEG and MEG signals, as well as the first to incorporate large-scale MEG pretraining. A total of 1,997 hours of EEG and 656 hours of MEG data are curated and standardised from publicly available sources for pretraining. Experiments show that BrainOmni outperforms both existing foundation models and state-of-the-art task-specific models on a range of downstream tasks. It also demonstrates strong generalisation to unseen EEG and MEG devices. Further analysis reveals that joint EEG-MEG (EMEG) training yields consistent improvements across both modalities. Code and checkpoints are publicly available at https://github.com/OpenTSLab/BrainOmni.

Siyin Wang, Chao-Han Huck Yang, Ji Wu et al.

Large language models (LLMs) can adapt to new tasks through in-context learning (ICL) based on a few examples presented in dialogue history without any model parameter update. Despite such convenience, the performance of ICL heavily depends on the quality of the in-context examples presented, which makes the in-context example selection approach a critical choice. This paper proposes a novel Bayesian in-Context example Selection method (ByCS) for ICL. Extending the inference probability conditioned on in-context examples based on Bayes' theorem, ByCS focuses on the inverse inference conditioned on test input. Following the assumption that accurate inverse inference probability (likelihood) will result in accurate inference probability (posterior), in-context examples are selected based on their inverse inference results. Diverse and extensive cross-tasking and cross-modality experiments are performed with speech, text, and image examples. Experimental results show the efficacy and robustness of our ByCS method on various models, tasks and modalities.

Wen Wu, Bo Li, Chao Zhang et al.

The subjective perception of emotion leads to inconsistent labels from human annotators. Typically, utterances lacking majority-agreed labels are excluded when training an emotion classifier, which cause problems when encountering ambiguous emotional expressions during testing. This paper investigates three methods to handle ambiguous emotion. First, we show that incorporating utterances without majority-agreed labels as an additional class in the classifier reduces the classification performance of the other emotion classes. Then, we propose detecting utterances with ambiguous emotions as out-of-domain samples by quantifying the uncertainty in emotion classification using evidential deep learning. This approach retains the classification accuracy while effectively detects ambiguous emotion expressions. Furthermore, to obtain fine-grained distinctions among ambiguous emotions, we propose representing emotion as a distribution instead of a single class label. The task is thus re-framed from classification to distribution estimation where every individual annotation is taken into account, not just the majority opinion. The evidential uncertainty measure is extended to quantify the uncertainty in emotion distribution estimation. Experimental results on the IEMOCAP and CREMA-D datasets demonstrate the superior capability of the proposed method in terms of majority class prediction, emotion distribution estimation, and uncertainty estimation.

Qiuming Zhao, Guangzhi Sun, Chao Zhang

Language diversity presents a significant challenge in speech-to-text (S2T) tasks, such as automatic speech recognition and translation. Traditional multi-lingual multi-task training approaches aim to address this by jointly optimising multiple speech recognition and translation tasks across various languages. While models like Whisper, built on these strategies, demonstrate strong performance, they still face issues of high computational cost, language interference, suboptimal training configurations, and limited extensibility. To overcome these challenges, we introduce LoRS-Merging (low-rank and sparse model merging), a novel technique designed to efficiently integrate models trained on different languages or tasks while preserving performance and reducing computational overhead. LoRS-Merging combines low-rank and sparse pruning to retain essential structures while eliminating redundant parameters, mitigating language interference, and enhancing extensibility. Experimental results across 10 languages demonstrate that LoRS-Merging significantly outperforms multi-lingual multi-task training, sequential training, and other merging methods, achieving over 20% improvement in normalised performance. Our findings suggest that model merging, particularly LoRS-Merging, is a scalable and effective complement to traditional multi-lingual training strategies for S2T applications.

Xiaodong Wu, Wenyi Yu, Chao Zhang et al.

Approximate Natural Gradient Descent (NGD) methods are an important family of optimisers for deep learning models, which use approximate Fisher information matrices to pre-condition gradients during training. The empirical Fisher (EF) method approximates the Fisher information matrix empirically by reusing the per-sample gradients collected during back-propagation. Despite its ease of implementation, the EF approximation has its theoretical and practical limitations. This paper investigates the inversely-scaled projection issue of EF, which is shown to be a major cause of its poor empirical approximation quality. An improved empirical Fisher (iEF) method is proposed to address this issue, which is motivated as a generalised NGD method from a loss reduction perspective, meanwhile retaining the practical convenience of EF. The exact iEF and EF methods are experimentally evaluated using practical deep learning setups. Optimisation experiments show that applying exact iEF directly as an optimiser provides strong convergence and generalisation. Additionally, under a novel empirical evaluation framework, the proposed iEF method shows consistently better approximation quality to exact Natural Gradient updates than both the EF and the more expensive sampled Fisher methods, meanwhile demonstrating the superior property of being robust to the choice of damping across tasks and training stages. Improving existing approximate NGD optimisers with iEF is expected to lead to better convergence and robustness. Furthermore, the iEF method also serves as a better approximation method to the Fisher information matrix itself, which enables the improvement of a variety of Fisher-based methods, not limited to the scope of optimisation.

Guangzhi Sun, Chao Zhang, Phil Woodland

The incorporation of biasing words obtained through contextual knowledge is of paramount importance in automatic speech recognition (ASR) applications. This paper proposes an innovative method for achieving end-to-end contextual ASR using graph neural network (GNN) encodings based on the tree-constrained pointer generator method. GNN node encodings facilitate lookahead for future word pieces in the process of ASR decoding at each tree node by incorporating information about all word pieces on the tree branches rooted from it. This results in a more precise prediction of the generation probability of the biasing words. The study explores three GNN encoding techniques, namely tree recursive neural networks, graph convolutional network (GCN), and GraphSAGE, along with different combinations of the complementary GCN and GraphSAGE structures. The performance of the systems was evaluated using the Librispeech and AMI corpus, following the visual-grounded contextual ASR pipeline. The findings indicate that using GNN encodings achieved consistent and significant reductions in word error rate (WER), particularly for words that are rare or have not been seen during the training process. Notably, the most effective combination of GNN encodings obtained more than 60% WER reduction for rare and unseen words compared to standard end-to-end systems.

Wen Wu, Chao Zhang, Philip C. Woodland

This paper proposes handling training data sparsity in speech-based automatic depression detection (SDD) using foundation models pre-trained with self-supervised learning (SSL). An analysis of SSL representations derived from different layers of pre-trained foundation models is first presented for SDD, which provides insight to suitable indicator for depression detection. Knowledge transfer is then performed from automatic speech recognition (ASR) and emotion recognition to SDD by fine-tuning the foundation models. Results show that the uses of oracle and ASR transcriptions yield similar SDD performance when the hidden representations of the ASR model is incorporated along with the ASR textual information. By integrating representations from multiple foundation models, state-of-the-art SDD results based on real ASR were achieved on the DAIC-WOZ dataset.

Zhiyun Lu, Yanwei Pan, Thibault Doutre et al.

End-to-end models have achieved state-of-the-art results on several automatic speech recognition tasks. However, they perform poorly when evaluated on long-form data, e.g., minutes long conversational telephony audio. One reason the model fails on long-form speech is that it has only seen short utterances during training. In this paper we study the effect of training utterance length on the word error rate (WER) for RNN-transducer (RNN-T) model. We compare two widely used training objectives, log loss (or RNN-T loss) and minimum word error rate (MWER) loss. We conduct experiments on telephony datasets in four languages. Our experiments show that for both losses, the WER on long-form speech reduces substantially as the training utterance length increases. The average relative WER gain is 15.7% for log loss and 8.8% for MWER loss. When training on short utterances, MWER loss leads to a lower WER than the log loss. Such difference between the two losses diminishes when the input length increases.

Guangzhi Sun, Chao Zhang, Philip C. Woodland

Contextual knowledge is important for real-world automatic speech recognition (ASR) applications. In this paper, a novel tree-constrained pointer generator (TCPGen) component is proposed that incorporates such knowledge as a list of biasing words into both attention-based encoder-decoder and transducer end-to-end ASR models in a neural-symbolic way. TCPGen structures the biasing words into an efficient prefix tree to serve as its symbolic input and creates a neural shortcut between the tree and the final ASR output distribution to facilitate recognising biasing words during decoding. Systems were trained and evaluated on the Librispeech corpus where biasing words were extracted at the scales of an utterance, a chapter, or a book to simulate different application scenarios. Experimental results showed that TCPGen consistently improved word error rates (WERs) compared to the baselines, and in particular, achieved significant WER reductions on the biasing words. TCPGen is highly efficient: it can handle 5,000 biasing words and distractors and only add a small overhead to memory use and computation cost.

Adnan Haider, Chao Zhang, Florian L. Kreyssig et al.

This paper presents a novel natural gradient and Hessian-free (NGHF) optimisation framework for neural network training that can operate efficiently in a distributed manner. It relies on the linear conjugate gradient (CG) algorithm to combine the natural gradient (NG) method with local curvature information from Hessian-free (HF) or other second-order methods. A solution to a numerical issue in CG allows effective parameter updates to be generated with far fewer CG iterations than usually used (e.g. 5-8 instead of 200). This work also presents a novel preconditioning approach to improve the progress made by individual CG iterations for models with shared parameters. Although applicable to other training losses and model structures, NGHF is investigated in this paper for lattice-based discriminative sequence training for hybrid hidden Markov model acoustic models using a standard recurrent neural network, long short-term memory, and time delay neural network models for output probability calculation. Automatic speech recognition experiments are reported on the multi-genre broadcast data set for a range of different acoustic model types. These experiments show that NGHF achieves larger word error rate reductions than standard stochastic gradient descent or Adam, while requiring orders of magnitude fewer parameter updates.

Wen Wu, Chao Zhang, Philip C. Woodland

In this paper, a novel two-branch neural network model structure is proposed for multimodal emotion recognition, which consists of a time synchronous branch (TSB) and a time asynchronous branch (TAB). To capture correlations between each word and its acoustic realisation, the TSB combines speech and text modalities at each input window frame and then does pooling across time to form a single embedding vector. The TAB, by contrast, provides cross-utterance information by integrating sentence text embeddings from a number of context utterances into another embedding vector. The final emotion classification uses both the TSB and the TAB embeddings. Experimental results on the IEMOCAP dataset demonstrate that the two-branch structure achieves state-of-the-art results in 4-way classification with all common test setups. When using automatic speech recognition (ASR) output instead of manually transcribed reference text, it is shown that the cross-utterance information considerably improves the robustness against ASR errors. Furthermore, by incorporating an extra class for all the other emotions, the final 5-way classification system with ASR hypotheses can be viewed as a prototype for more realistic emotion recognition systems.

Guangzhi Sun, Chao Zhang, Phil Woodland

Significant progress has recently been made in speaker diarisation after the introduction of d-vectors as speaker embeddings extracted from neural network (NN) speaker classifiers for clustering speech segments. To extract better-performing and more robust speaker embeddings, this paper proposes a c-vector method by combining multiple sets of complementary d-vectors derived from systems with different NN components. Three structures are used to implement the c-vectors, namely 2D self-attentive, gated additive, and bilinear pooling structures, relying on attention mechanisms, a gating mechanism, and a low-rank bilinear pooling mechanism respectively. Furthermore, a neural-based single-pass speaker diarisation pipeline is also proposed in this paper, which uses NNs to achieve voice activity detection, speaker change point detection, and speaker embedding extraction. Experiments and detailed analyses are conducted on the challenging AMI and NIST RT05 datasets which consist of real meetings with 4--10 speakers and a wide range of acoustic conditions. For systems trained on the AMI training set, relative speaker error rate (SER) reductions of 13% and 29% are obtained by using c-vectors instead of d-vectors on the AMI dev and eval sets respectively, and a relative reduction of 15% in SER is observed on RT05, which shows the robustness of the proposed methods. By incorporating VoxCeleb data into the training set, the best c-vector system achieved 7%, 17% and16% relative SER reduction compared to the d-vector on the AMI dev, eval, and RT05 sets respectively

G. Sun, C. Zhang, P. C. Woodland

The effective exploitation of richer contextual information in language models (LMs) is a long-standing research problem for automatic speech recognition (ASR). A cross-utterance LM (CULM) is proposed in this paper, which augments the input to a standard long short-term memory (LSTM) LM with a context vector derived from past and future utterances using an extraction network. The extraction network uses another LSTM to encode surrounding utterances into vectors which are integrated into a context vector using either a projection of LSTM final hidden states, or a multi-head self-attentive layer. In addition, an acoustic error sampling technique is proposed to reduce the mismatch between training and test-time. This is achieved by considering possible ASR errors into the model training procedure, and can therefore improve the word error rate (WER). Experiments performed on both AMI and Switchboard datasets show that CULMs outperform the LSTM LM baseline WER. In particular, the CULM with a self-attentive layer-based extraction network and acoustic error sampling achieves 0.6% absolute WER reduction on AMI, 0.3% WER reduction on the Switchboard part and 0.9% WER reduction on the Callhome part of Eval2000 test set over the respective baselines.

Chao Zhang, Zichao Yang, Xiaodong He et al.

Deep learning methods have revolutionized speech recognition, image recognition, and natural language processing since 2010. Each of these tasks involves a single modality in their input signals. However, many applications in the artificial intelligence field involve multiple modalities. Therefore, it is of broad interest to study the more difficult and complex problem of modeling and learning across multiple modalities. In this paper, we provide a technical review of available models and learning methods for multimodal intelligence. The main focus of this review is the combination of vision and natural language modalities, which has become an important topic in both the computer vision and natural language processing research communities. This review provides a comprehensive analysis of recent works on multimodal deep learning from three perspectives: learning multimodal representations, fusing multimodal signals at various levels, and multimodal applications. Regarding multimodal representation learning, we review the key concepts of embedding, which unify multimodal signals into a single vector space and thereby enable cross-modality signal processing. We also review the properties of many types of embeddings that are constructed and learned for general downstream tasks. Regarding multimodal fusion, this review focuses on special architectures for the integration of representations of unimodal signals for a particular task. Regarding applications, selected areas of a broad interest in the current literature are covered, including image-to-text caption generation, text-to-image generation, and visual question answering. We believe that this review will facilitate future studies in the emerging field of multimodal intelligence for related communities.

Qiujia Li, Florian L. Kreyssig, Chao Zhang et al.

In this paper, we propose Discriminative Neural Clustering (DNC) that formulates data clustering with a maximum number of clusters as a supervised sequence-to-sequence learning problem. Compared to traditional unsupervised clustering algorithms, DNC learns clustering patterns from training data without requiring an explicit definition of a similarity measure. An implementation of DNC based on the Transformer architecture is shown to be effective on a speaker diarisation task using the challenging AMI dataset. Since AMI contains only 147 complete meetings as individual input sequences, data scarcity is a significant issue for training a Transformer model for DNC. Accordingly, this paper proposes three data augmentation schemes: sub-sequence randomisation, input vector randomisation, and Diaconis augmentation, which generates new data samples by rotating the entire input sequence of L2-normalised speaker embeddings. Experimental results on AMI show that DNC achieves a reduction in speaker error rate (SER) of 29.4% relative to spectral clustering.

Guangzhi Sun, Chao Zhang, Phil Woodland

Speaker diarisation systems often cluster audio segments using speaker embeddings such as i-vectors and d-vectors. Since different types of embeddings are often complementary, this paper proposes a generic framework to improve performance by combining them into a single embedding, referred to as a c-vector. This combination uses a 2-dimensional (2D) self-attentive structure, which extends the standard self-attentive layer by averaging not only across time but also across different types of embeddings. Two types of 2D self-attentive structure in this paper are the simultaneous combination and the consecutive combination, adopting a single and multiple self-attentive layers respectively. The penalty term in the original self-attentive layer which is jointly minimised with the objective function to encourage diversity of annotation vectors is also modified to obtain not only different local peaks but also the overall trends in the multiple annotation vectors. Experiments on the AMI meeting corpus show that our modified penalty term improves the d- vector relative speaker error rate (SER) by 6% and 21% for d-vector systems, and a 10% further relative SER reduction can be obtained using the c-vector from our best 2D self-attentive structure.