SA-WavLM: Speaker-Aware Self-Supervised Pre-training for Mixture Speech

It was shown that pre-trained models with self-supervised learning (SSL) techniques are effective in various downstream speech tasks. However, most such models are trained on single-speaker speech data, limiting their effectiveness in mixture speech. This motivates us to explore pre-training on mixture speech. This work presents SA-WavLM, a novel pre-trained model for mixture speech. Specifically, SA-WavLM follows an "extract-merge-predict" pipeline in which the representations of each speaker in the input mixture are first extracted individually and then merged before the final prediction. In this pipeline, SA-WavLM performs speaker-informed extractions with the consideration of the interactions between different speakers. Furthermore, a speaker shuffling strategy is proposed to enhance the robustness towards the speaker absence. Experiments show that SA-WavLM either matches or improves upon the state-of-the-art pre-trained models.

SD-Eval: A Benchmark Dataset for Spoken Dialogue Understanding Beyond Words

Speech encompasses a wealth of information, including but not limited to content, paralinguistic, and environmental information. This comprehensive nature of speech significantly impacts communication and is crucial for human-computer interaction. Chat-Oriented Large Language Models (LLMs), known for their general-purpose assistance capabilities, have evolved to handle multi-modal inputs, including speech. Although these models can be adept at recognizing and analyzing speech, they often fall short of generating appropriate responses. We argue that this is due to the lack of principles on task definition and model development, which requires open-source datasets and metrics suitable for model evaluation. To bridge the gap, we present SD-Eval, a benchmark dataset aimed at multidimensional evaluation of spoken dialogue understanding and generation. SD-Eval focuses on paralinguistic and environmental information and includes 7,303 utterances, amounting to 8.76 hours of speech data. The data is aggregated from eight public datasets, representing four perspectives: emotion, accent, age, and background sound. To assess the SD-Eval benchmark dataset, we implement three different models and construct a training set following a similar process as SD-Eval. The training set contains 1,052.72 hours of speech data and 724.4k utterances. We also conduct a comprehensive evaluation using objective evaluation methods (e.g. BLEU and ROUGE), subjective evaluations and LLM-based metrics for the generated responses. Models conditioned with paralinguistic and environmental information outperform their counterparts in both objective and subjective measures. Moreover, experiments demonstrate LLM-based metrics show a higher correlation with human evaluation compared to traditional metrics. We open-source SD-Eval at https://github.com/amphionspace/SD-Eval.

Text-guided HuBERT: Self-Supervised Speech Pre-training via Generative Adversarial Networks

Human language can be expressed in either written or spoken form, i.e. text or speech. Humans can acquire knowledge from text to improve speaking and listening. However, the quest for speech pre-trained models to leverage unpaired text has just started. In this paper, we investigate a new way to pre-train such a joint speech-text model to learn enhanced speech representations and benefit various speech-related downstream tasks. Specifically, we propose a novel pre-training method, text-guided HuBERT, or T-HuBERT, which performs self-supervised learning over speech to derive phoneme-like discrete representations. And these phoneme-like pseudo-label sequences are firstly derived from speech via the generative adversarial networks (GAN) to be statistically similar to those from additional unpaired textual data. In this way, we build a bridge between unpaired speech and text in an unsupervised manner. Extensive experiments demonstrate the significant superiority of our proposed method over various strong baselines, which achieves up to 15.3% relative Word Error Rate (WER) reduction on the LibriSpeech dataset.

The NUS-HLT System for ICASSP2024 ICMC-ASR Grand Challenge

This paper summarizes our team's efforts in both tracks of the ICMC-ASR Challenge for in-car multi-channel automatic speech recognition. Our submitted systems for ICMC-ASR Challenge include the multi-channel front-end enhancement and diarization, training data augmentation, speech recognition modeling with multi-channel branches. Tested on the offical Eval1 and Eval2 set, our best system achieves a relative 34.3% improvement in CER and 56.5% improvement in cpCER, compared to the offical baseline system.

USED: Universal Speaker Extraction and Diarization

Speaker extraction and diarization are two crucial enabling techniques for speech applications. Speaker extraction aims to extract a target speaker's voice from a multi-talk mixture, while speaker diarization demarcates speech segments by speaker, identifying `who spoke when'. The previous studies have typically treated the two tasks independently. However, the two tasks share a similar objective, that is to disentangle the speakers in the spectral domain for the former but in the temporal domain for the latter. It is logical to believe that the speaker turns obtained from speaker diarization can benefit speaker extraction, while the extracted speech offers more accurate speaker turns than the mixture speech. In this paper, we propose a unified framework called Universal Speaker Extraction and Diarization (USED). We extend the existing speaker extraction model to simultaneously extract the waveforms of all speakers. We also employ a scenario-aware differentiated loss function to address the problem of sparsely overlapped speech in real-world conversations. We show that the USED model significantly outperforms the baselines for both speaker extraction and diarization tasks, in both highly overlapped and sparsely overlapped scenarios. Audio samples are available at https://ajyy.github.io/demo/USED/.

Self-Supervised Acoustic Word Embedding Learning via Correspondence Transformer Encoder

Acoustic word embeddings (AWEs) aims to map a variable-length speech segment into a fixed-dimensional representation. High-quality AWEs should be invariant to variations, such as duration, pitch and speaker. In this paper, we introduce a novel self-supervised method to learn robust AWEs from a large-scale unlabelled speech corpus. Our model, named Correspondence Transformer Encoder (CTE), employs a teacher-student learning framework. We train the model based on the idea that different realisations of the same word should be close in the underlying embedding space. Specifically, we feed the teacher and student encoder with different acoustic instances of the same word and pre-train the model with a word-level loss. Our experiments show that the embeddings extracted from the proposed CTE model are robust to speech variations, e.g. speakers and domains. Additionally, when evaluated on Xitsonga, a low-resource cross-lingual setting, the CTE model achieves new state-of-the-art performance.

token2vec: A Joint Self-Supervised Pre-training Framework Using Unpaired Speech and Text

Self-supervised pre-training has been successful in both text and speech processing. Speech and text offer different but complementary information. The question is whether we are able to perform a speech-text joint pre-training on unpaired speech and text. In this paper, we take the idea of self-supervised pre-training one step further and propose token2vec, a novel joint pre-training framework for unpaired speech and text based on discrete representations of speech. Firstly, due to the distinct characteristics between speech and text modalities, where speech is continuous while text is discrete, we first discretize speech into a sequence of discrete speech tokens to solve the modality mismatch problem. Secondly, to solve the length mismatch problem, where the speech sequence is usually much longer than text sequence, we convert the words of text into phoneme sequences and randomly repeat each phoneme in the sequences. Finally, we feed the discrete speech and text tokens into a modality-agnostic Transformer encoder and pre-train with token-level masking language modeling (tMLM). Experiments show that token2vec is significantly superior to various speech-only pre-training baselines, with up to 17.7% relative WER reduction. Token2vec model is also validated on a non-ASR task, i.e., spoken intent classification, and shows good transferability.

CoBERT: Self-Supervised Speech Representation Learning Through Code Representation Learning

Speech is the surface form of a finite set of phonetic units, which can be represented by discrete codes. We propose the Code BERT (CoBERT) approach for self-supervised speech representation learning. The idea is to convert an utterance to a sequence of discrete codes, and perform code representation learning, where we predict the code representations based on a masked view of the original speech input. Unlike the prior self-distillation approaches of which the teacher and the student are of the same modality, our target model predicts representations from a different modality. CoBERT outperforms the most recent state-of-the-art performance on the ASR task and brings significant improvements on the SUPERB speech translation (ST) task.

SpeechUT: Bridging Speech and Text with Hidden-Unit for Encoder-Decoder Based Speech-Text Pre-training

The rapid development of single-modal pre-training has prompted researchers to pay more attention to cross-modal pre-training methods. In this paper, we propose a unified-modal speech-unit-text pre-training model, SpeechUT, to connect the representations of a speech encoder and a text decoder with a shared unit encoder. Leveraging hidden-unit as an interface to align speech and text, we can decompose the speech-to-text model into a speech-to-unit model and a unit-to-text model, which can be jointly pre-trained with unpaired speech and text data respectively. Our proposed SpeechUT is fine-tuned and evaluated on automatic speech recognition (ASR) and speech translation (ST) tasks. Experimental results show that SpeechUT gets substantial improvements over strong baselines, and achieves state-of-the-art performance on both the LibriSpeech ASR and MuST-C ST tasks. To better understand the proposed SpeechUT, detailed analyses are conducted. The code and pre-trained models are available at https://aka.ms/SpeechUT.

The YiTrans End-to-End Speech Translation System for IWSLT 2022 Offline Shared Task

This paper describes the submission of our end-to-end YiTrans speech translation system for the IWSLT 2022 offline task, which translates from English audio to German, Chinese, and Japanese. The YiTrans system is built on large-scale pre-trained encoder-decoder models. More specifically, we first design a multi-stage pre-training strategy to build a multi-modality model with a large amount of labeled and unlabeled data. We then fine-tune the corresponding components of the model for the downstream speech translation tasks. Moreover, we make various efforts to improve performance, such as data filtering, data augmentation, speech segmentation, model ensemble, and so on. Experimental results show that our YiTrans system obtains a significant improvement than the strong baseline on three translation directions, and it achieves +5.2 BLEU improvements over last year's optimal end-to-end system on tst2021 English-German. Our final submissions rank first on English-German and English-Chinese end-to-end systems in terms of the automatic evaluation metric. We make our code and models publicly available.