Berlin Chen

Yi-Cheng Wang, Li-Ting Pai, Bi-Cheng Yan et al.

End-to-end (E2E) automatic speech recognition (ASR) models have become standard practice for various commercial applications. However, in real-world scenarios, the long-tailed nature of word distribution often leads E2E ASR models to perform well on common words but fall short in recognizing uncommon ones. Recently, the notion of a contextual adapter (CA) was proposed to infuse external knowledge represented by a context word list into E2E ASR models. Although CA can improve recognition performance on rare words, two crucial data imbalance problems remain. First, when using low-frequency words as context words during training, since these words rarely occur in the utterance, CA becomes prone to overfit on attending to the <no-context> token due to higher-frequency words not being present in the context list. Second, the long-tailed distribution within the context list itself still causes the model to perform poorly on low-frequency context words. In light of this, we explore in-depth the impact of altering the context list to have words with different frequency distributions on model performance, and meanwhile extend CA with a simple yet effective context-balanced learning objective. A series of experiments conducted on the AISHELL-1 benchmark dataset suggests that using all vocabulary words from the training corpus as the context list and pairing them with our balanced objective yields the best performance, demonstrating a significant reduction in character error rate (CER) by up to 1.21% and a more pronounced 9.44% reduction in the error rate of zero-shot words.

Tzu-Ting Yang, Hsin-Wei Wang, Yi-Cheng Wang et al.

With the massive developments of end-to-end (E2E) neural networks, recent years have witnessed unprecedented breakthroughs in automatic speech recognition (ASR). However, the codeswitching phenomenon remains a major obstacle that hinders ASR from perfection, as the lack of labeled data and the variations between languages often lead to degradation of ASR performance. In this paper, we focus exclusively on improving the acoustic encoder of E2E ASR to tackle the challenge caused by the codeswitching phenomenon. Our main contributions are threefold: First, we introduce a novel disentanglement loss to enable the lower-layer of the encoder to capture inter-lingual acoustic information while mitigating linguistic confusion at the higher-layer of the encoder. Second, through comprehensive experiments, we verify that our proposed method outperforms the prior-art methods using pretrained dual-encoders, meanwhile having access only to the codeswitching corpus and consuming half of the parameterization. Third, the apparent differentiation of the encoders' output features also corroborates the complementarity between the disentanglement loss and the mixture-of-experts (MoE) architecture.

Cheng-Yeh Yang, Kuan-Tang Huang, Chien-Chun Wang et al.

A pooling mechanism is essential for mean opinion score (MOS) prediction, facilitating the transformation of variable-length audio features into a concise fixed-size representation that effectively encodes speech quality. Existing pooling methods typically operate at a singular granularity, concentrating either on a comprehensive global perspective or a detailed frame-level analysis, which may overlook complementary perceptual insights. To address this limitation, we introduce the Dual-Resolution Attentive Statistics Pooling (DRASP) framework. DRASP integrates both coarse-grained, global statistical summaries and fine-grained, attentive analyses of perceptually significant segments. This dual-view architecture empowers our model to formulate a more thorough and robust representation, capturing both the overarching structural context and salient local details concurrently. Extensive experiments validate the effectiveness and strong generalization ability of the proposed framework. It consistently outperforms various baseline methods across diverse datasets (MusicEval and AES-Natural), MOS prediction backbones (including a CLAP-based model and AudioBox-Aesthetics), and different audio generation systems, achieving a relative improvement of 10.39% in system-level Spearman's rank correlation coefficient (SRCC) over the widely-used average pooling approach.

Chien-Chun Wang, Kuan-Tang Huang, Cheng-Yeh Yang et al.

Evaluating audio generation systems, including text-to-music (TTM), text-to-speech (TTS), and text-to-audio (TTA), remains challenging due to the subjective and multi-dimensional nature of human perception. Existing methods treat mean opinion score (MOS) prediction as a regression problem, but standard regression losses overlook the relativity of perceptual judgments. To address this limitation, we introduce QAMRO, a novel Quality-aware Adaptive Margin Ranking Optimization framework that seamlessly integrates regression objectives from different perspectives, aiming to highlight perceptual differences and prioritize accurate ratings. Our framework leverages pre-trained audio-text models such as CLAP and Audiobox-Aesthetics, and is trained exclusively on the official AudioMOS Challenge 2025 dataset. It demonstrates superior alignment with human evaluations across all dimensions, significantly outperforming robust baseline models.

Bi-Cheng Yan, Hsin-Wei Wang, Shih-Hsuan Chiu et al.

Conversational speech normally is embodied with loose syntactic structures at the utterance level but simultaneously exhibits topical coherence relations across consecutive utterances. Prior work has shown that capturing longer context information with a recurrent neural network or long short-term memory language model (LM) may suffer from the recent bias while excluding the long-range context. In order to capture the long-term semantic interactions among words and across utterances, we put forward disparate conversation history fusion methods for language modeling in automatic speech recognition (ASR) of conversational speech. Furthermore, a novel audio-fusion mechanism is introduced, which manages to fuse and utilize the acoustic embeddings of a current utterance and the semantic content of its corresponding conversation history in a cooperative way. To flesh out our ideas, we frame the ASR N-best hypothesis rescoring task as a prediction problem, leveraging BERT, an iconic pre-trained LM, as the ingredient vehicle to facilitate selection of the oracle hypothesis from a given N-best hypothesis list. Empirical experiments conducted on the AMI benchmark dataset seem to demonstrate the feasibility and efficacy of our methods in relation to some current top-of-line methods. The proposed methods not only achieve significant inference time reduction but also improve the ASR performance for conversational speech.

Fu-An Chao, Jeih-weih Hung, Berlin Chen

In recent decades, many studies have suggested that phase information is crucial for speech enhancement (SE), and time-domain single-channel speech enhancement techniques have shown promise in noise suppression and robust automatic speech recognition (ASR). This paper presents a continuation of the above lines of research and explores two effective SE methods that consider phase information in time domain and frequency domain of speech signals, respectively. Going one step further, we put forward a novel cross-domain speech enhancement model and a bi-projection fusion (BPF) mechanism for noise-robust ASR. To evaluate the effectiveness of our proposed method, we conduct an extensive set of experiments on the publicly-available Aishell-1 Mandarin benchmark speech corpus. The evaluation results confirm the superiority of our proposed method in relation to a few current top-of-the-line time-domain and frequency-domain SE methods in both enhancement and ASR evaluation metrics for the test set of scenarios contaminated with seen and unseen noise, respectively.

Fu-An Chao, Shao-Wei Fan Jiang, Bi-Cheng Yan et al.

Due to the unprecedented breakthroughs brought about by deep learning, speech enhancement (SE) techniques have been developed rapidly and play an important role prior to acoustic modeling to mitigate noise effects on speech. To increase the perceptual quality of speech, current state-of-the-art in the SE field adopts adversarial training by connecting an objective metric to the discriminator. However, there is no guarantee that optimizing the perceptual quality of speech will necessarily lead to improved automatic speech recognition (ASR) performance. In this study, we present TENET, a novel Time-reversal Enhancement NETwork, which leverages the transformation of an input noisy signal itself, i.e., the time-reversed version, in conjunction with the siamese network and complex dual-path transformer to promote SE performance for noise-robust ASR. Extensive experiments conducted on the Voicebank-DEMAND dataset show that TENET can achieve state-of-the-art results compared to a few top-of-the-line methods in terms of both SE and ASR evaluation metrics. To demonstrate the model generalization ability, we further evaluate TENET on the test set of scenarios contaminated with unseen noise, and the results also confirm the superiority of this promising method.

Kuan-Yu Chen, Shih-Hung Liu, Berlin Chen et al.

In the context of natural language processing, representation learning has emerged as a newly active research subject because of its excellent performance in many applications. Learning representations of words is a pioneering study in this school of research. However, paragraph (or sentence and document) embedding learning is more suitable/reasonable for some tasks, such as sentiment classification and document summarization. Nevertheless, as far as we are aware, there is relatively less work focusing on the development of unsupervised paragraph embedding methods. Classic paragraph embedding methods infer the representation of a given paragraph by considering all of the words occurring in the paragraph. Consequently, those stop or function words that occur frequently may mislead the embedding learning process to produce a misty paragraph representation. Motivated by these observations, our major contributions in this paper are twofold. First, we propose a novel unsupervised paragraph embedding method, named the essence vector (EV) model, which aims at not only distilling the most representative information from a paragraph but also excluding the general background information to produce a more informative low-dimensional vector representation for the paragraph. Second, in view of the increasing importance of spoken content processing, an extension of the EV model, named the denoising essence vector (D-EV) model, is proposed. The D-EV model not only inherits the advantages of the EV model but also can infer a more robust representation for a given spoken paragraph against imperfect speech recognition.