Search

Language-queried target sound extraction (TSE) aims to extract specific sounds from mixtures based on language queries. Traditional fully-supervised training schemes require extensively annotated parallel audio-text data, which are labor-intensive. We introduce a language-free training scheme, requiring only unlabelled audio clips for TSE model training by utilizing the multi-modal representation alignment nature of the contrastive language-audio pre-trained model (CLAP). In a vanilla language-free training stage, target audio is encoded using the pre-trained CLAP audio encoder to form a condition embedding for the TSE model, while during inference, user language queries are encoded by CLAP text encoder. This straightforward approach faces challenges due to the modality gap between training and inference queries and information leakage from direct exposure to target audio during training. To address this, we propose a retrieval-augmented strategy. Specifically, we create an embedding cache using audio captions generated by a large language model (LLM). During training, target audio embeddings retrieve text embeddings from this cache to use as condition embeddings, ensuring consistent modalities between training and inference and eliminating information leakage. Extensive experiment results show that our retrieval-augmented approach achieves consistent and notable performance improvements over existing state-of-the-art with better generalizability., Comment: Submitted to ICASSP 2025

Target-speaker automatic speech recognition (ASR) aims to transcribe the desired speech of a target speaker from multi-talker overlapped utterances. Most of the existing target-speaker ASR (TS-ASR) methods involve either training from scratch or fully fine-tuning a pre-trained model, leading to significant training costs and becoming inapplicable to large foundation models. This work leverages prompt tuning, a parameter-efficient fine-tuning approach, to extend Whisper, a large-scale single-talker ASR model, to TS-ASR. Variants of prompt tuning approaches along with their configurations are explored and optimized for TS-ASR.Experimental results show that prompt tuning can achieve performance comparable to state-of-the-art full training approaches while only requiring about 1\% of task-specific model parameters. Notably, the original Whisper's features, such as inverse text normalization and timestamp tagging, are retained in target-speaker ASR, keeping the generated transcriptions natural and informative., Comment: ICASSP 2024

Transformer-based speech recognition (ASR) model with deep layers exhibited significant performance improvement. However, the model is inefficient for deployment on resource-constrained devices. Layer pruning (LP) is a commonly used compression method to remove redundant layers. Previous studies on LP usually identify the redundant layers according to a task-specific evaluation metric. They are time-consuming for models with a large number of layers, even in a greedy search manner. To address this problem, we propose CoMFLP, a fast search LP algorithm based on correlation measure. The correlation between layers is computed to generate a correlation matrix, which identifies the redundancy among layers. The search process is carried out in two steps: (1) coarse search: to determine top $K$ candidates by pruning the most redundant layers based on the correlation matrix; (2) fine search: to select the best pruning proposal among $K$ candidates using a task-specific evaluation metric. Experiments on an ASR task show that the pruning proposal determined by CoMFLP outperforms existing LP methods while only requiring constant time complexity. The code is publicly available at https://github.com/louislau1129/CoMFLP., Comment: Accepted by Interspeech 2023

Whisper is a powerful automatic speech recognition (ASR) model. Nevertheless, its zero-shot performance on low-resource speech requires further improvement. Child speech, as a representative type of low-resource speech, is leveraged for adaptation. Recently, parameter-efficient fine-tuning (PEFT) in NLP was shown to be comparable and even better than full fine-tuning, while only needing to tune a small set of trainable parameters. However, current PEFT methods have not been well examined for their effectiveness on Whisper. In this paper, only parameter composition types of PEFT approaches such as LoRA and Bitfit are investigated as they do not bring extra inference costs. Different popular PEFT methods are examined. Particularly, we compare LoRA and AdaLoRA and figure out the learnable rank coefficient is a good design. Inspired by the sparse rank distribution allocated by AdaLoRA, a novel PEFT approach Sparsely Shared LoRA (S2-LoRA) is proposed. The two low-rank decomposed matrices are globally shared. Each weight matrix only has to maintain its specific rank coefficients that are constrained to be sparse. Experiments on low-resource Chinese child speech show that with much fewer trainable parameters, S2-LoRA can achieve comparable in-domain adaptation performance to AdaLoRA and exhibit better generalization ability on out-of-domain data. In addition, the rank distribution automatically learned by S2-LoRA is found to have similar patterns to AdaLoRA's allocation., Comment: Accepted by ICASSP 2024

Very deep models for speaker recognition (SR) have demonstrated remarkable performance improvement in recent research. However, it is impractical to deploy these models for on-device applications with constrained computational resources. On the other hand, light-weight models are highly desired in practice despite their sub-optimal performance. This research aims to improve light-weight SR models through large-scale label-free knowledge distillation (KD). Existing KD approaches for SR typically require speaker labels to learn task-specific knowledge, due to the inefficiency of conventional loss for distillation. To address the inefficiency problem and achieve label-free KD, we propose to employ the contrastive loss from self-supervised learning for distillation. Extensive experiments are conducted on a collection of public speech datasets from diverse sources. Results on light-weight SR models show that the proposed approach of label-free KD with contrastive loss consistently outperforms both conventional distillation methods and self-supervised learning methods by a significant margin.

Probabilistic linear discriminant analysis (PLDA) is commonly used in speaker verification systems to score the similarity of speaker embeddings. Recent studies improved the performance of PLDA in domain-matched conditions by diagonalizing its covariance. We suspect such brutal pruning approach could eliminate its capacity in modeling dimension correlation of speaker embeddings, leading to inadequate performance with domain adaptation. This paper explores two alternative covariance regularization approaches, namely, interpolated PLDA and sparse PLDA, to tackle the problem. The interpolated PLDA incorporates the prior knowledge from cosine scoring to interpolate the covariance of PLDA. The sparse PLDA introduces a sparsity penalty to update the covariance. Experimental results demonstrate that both approaches outperform diagonal regularization noticeably with domain adaptation. In addition, in-domain data can be significantly reduced when training sparse PLDA for domain adaptation.

State-of-art speaker verification (SV) systems use a back-end model to score the similarity of speaker embeddings extracted from a neural network model. The commonly used back-end models are the cosine scoring and the probabilistic linear discriminant analysis (PLDA) scoring. With the recently developed neural embeddings, the theoretically more appealing PLDA approach is found to have no advantage against or even be inferior the simple cosine scoring in terms of SV system performance. This paper presents an investigation on the relation between the two scoring approaches, aiming to explain the above counter-intuitive observation. It is shown that the cosine scoring is essentially a special case of PLDA scoring. In other words, by properly setting the parameters of PLDA, the two back-ends become equivalent. As a consequence, the cosine scoring not only inherits the basic assumptions for the PLDA but also introduces additional assumptions on the properties of input embeddings. Experiments show that the dimensional independence assumption required by the cosine scoring contributes most to the performance gap between the two methods under the domain-matched condition. When there is severe domain mismatch and the dimensional independence assumption does not hold, the PLDA would perform better than the cosine for domain adaptation., Comment: submitted to interspeech2022

This technical report describes our submission to the 2021 SLT Children Speech Recognition Challenge (CSRC) Track 1. Our approach combines the use of a joint CTC-attention end-to-end (E2E) speech recognition framework, transfer learning, data augmentation and development of various language models. Procedures of data pre-processing, the background and the course of system development are described. The analysis of the experiment results, as well as the comparison between the E2E and DNN-HMM hybrid system are discussed in detail. Our system achieved a character error rate (CER) of 20.1% in our designated test set, and 23.6% in the official evaluation set, which is placed at 10-th overall., Comment: Submitted to 2021 SLT Children Speech Recognition Challenge (CSRC)

Speech signal is constituted and contributed by various informative factors, such as linguistic content and speaker characteristic. There have been notable recent studies attempting to factorize speech signal into these individual factors without requiring any annotation. These studies typically assume continuous representation for linguistic content, which is not in accordance with general linguistic knowledge and may make the extraction of speaker information less successful. This paper proposes the mixture factorized auto-encoder (mFAE) for unsupervised deep factorization. The encoder part of mFAE comprises a frame tokenizer and an utterance embedder. The frame tokenizer models linguistic content of input speech with a discrete categorical distribution. It performs frame clustering by assigning each frame a soft mixture label. The utterance embedder generates an utterance-level vector representation. A frame decoder serves to reconstruct speech features from the encoders'outputs. The mFAE is evaluated on speaker verification (SV) task and unsupervised subword modeling (USM) task. The SV experiments on VoxCeleb 1 show that the utterance embedder is capable of extracting speaker-discriminative embeddings with performance comparable to a x-vector baseline. The USM experiments on ZeroSpeech 2017 dataset verify that the frame tokenizer is able to capture linguistic content and the utterance embedder can acquire speaker-related information.

This study addresses the problem of unsupervised subword unit discovery from untranscribed speech. It forms the basis of the ultimate goal of ZeroSpeech 2019, building text-to-speech systems without text labels. In this work, unit discovery is formulated as a pipeline of phonetically discriminative feature learning and unit inference. One major difficulty in robust unsupervised feature learning is dealing with speaker variation. Here the robustness towards speaker variation is achieved by applying adversarial training and FHVAE based disentangled speech representation learning. A comparison of the two approaches as well as their combination is studied in a DNN-bottleneck feature (DNN-BNF) architecture. Experiments are conducted on ZeroSpeech 2019 and 2017. Experimental results on ZeroSpeech 2017 show that both approaches are effective while the latter is more prominent, and that their combination brings further marginal improvement in across-speaker condition. Results on ZeroSpeech 2019 show that in the ABX discriminability task, our approaches significantly outperform the official baseline, and are competitive to or even outperform the official topline. The proposed unit sequence smoothing algorithm improves synthesis quality, at a cost of slight decrease in ABX discriminability., Comment: 5 pages, 3 figures, accepted for publication in INTERSPEECH 2019, Graz, Austria