Huang-Cheng Chou
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Huang-Cheng Chou has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Ali N. Salman
The University of Texas at Dallas, Richardson, TX, USA
4 shared publications - Carlos Busso
Language Technologies Institute, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA
4 shared publications - Lucas Goncalves
The University of Texas at Dallas, Richardson, TX, USA
4 shared publications - Chi-Chun Lee
Department of Electrical Engineering, National Tsing Hua University, Hsinchu, Taiwan
3 shared publications - Seong-Gyun Leem
Reality Labs at Meta Platforms, Inc., USA
2 shared publications - Abinay Reddy Naini
Erik Jonsson School of Engineering and Computer Science, University of Texas at Dallas, Richardson, TX, USA
1 shared publication
Latest Publications
- Open Access
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. PP (3) , 1-19
The MSP-Podcast Corpus
The availability of large, high-quality emotional speech databases is essential for advancing speech emotion recognition (SER) in real-world scenarios. However, many existing databases face limitations in size, emotional balance, and speaker diversity. This study describes the MSP-Podcast corpus, summarizing our ten-year effort. The corpus consists of over 400 hours of diverse audio samples from various audio sharing websites, all of which have Common Licenses that permit the distribution of the corpus. We annotate the corpus with rich emotional labels, including primary (single dominant emotion) and secondary (multiple emotions perceived in the audio) emotional categories, as well as emotional attributes for valence, arousal, and dominance. At least five raters annotate these emotional labels. The corpus also has speaker identification for most samples, and human transcriptions of the lexical content of the sentences for the entire corpus. The data collection protocol includes a machine learning-driven pipeline for selecting emotionally diverse recordings, ensuring a balanced and varied representation of emotions across speakers and environments. The resulting database provides a comprehensive, high-quality resource, better suited for advancing SER systems in practical, real-world scenarios.
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. PP (3) , 1-19
Open AccessThe MSP-Podcast Corpus
The availability of large, high-quality emotional speech databases is essential for advancing speech emotion recognition (SER) in real-world scenarios. However, many existing databases face limitations in size, emotional balance, and speaker diversity. This study describes the MSP-Podcast corpus, summarizing our ten-year effort. The corpus consists of over 400 hours of diverse audio samples from various audio sharing websites, all of which have Common Licenses that permit the distribution of the corpus. We annotate the corpus with rich emotional labels, including primary (single dominant emotion) and secondary (multiple emotions perceived in the audio) emotional categories, as well as emotional attributes for valence, arousal, and dominance. At least five raters annotate these emotional labels. The corpus also has speaker identification for most samples, and human transcriptions of the lexical content of the sentences for the entire corpus. The data collection protocol includes a machine learning-driven pipeline for selecting emotionally diverse recordings, ensuring a balanced and varied representation of emotions across speakers and environments. The resulting database provides a comprehensive, high-quality resource, better suited for advancing SER systems in practical, real-world scenarios.
- Top Cited Papers
- 19 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Audio, Speech and Language Processing
- Vol. 34, 2062-2076
DeSTA2.5-Audio: Toward General-Purpose Large Audio Language Model With Self-Generated Cross-Modal Alignment
We introduce DeSTA2.5-Audio, a general-purpose Large Audio Language Model (LALM) designed for robust auditory perception and instruction-following. Recent LALMs augment Large Language Models (LLMs) with auditory capabilities by training on large-scale audio-instruction datasets. However, existing LALMs have often suffered from the catastrophic forgetting of the LLM's original abilities. Therefore, balancing knowledge retention and audio perception has become a critical challenge. To address this, we revisit the data construction pipeline and propose a self-generated cross-modal alignment strategy in which the backbone LLM generates its own training targets, named DeSTA. This approach aims at preserving the LLM's native language proficiency thereby enabling zero-shot generalization without task-specific tuning. We construct DeSTA-AQA5M, a large-scale, task-agnostic dataset containing 5 million training samples derived from 7,000 hours of audio spanning 50 diverse datasets, including speech, environmental sounds, and music. DeSTA2.5-Audio achieves state-of-the-art or competitive performance across a wide range of audio-language benchmarks, including Dynamic-SUPERB, MMAU, SAKURA, Speech-IFEval, and VoiceBench. Comprehensive comparative studies demonstrate that our self-generated strategy outperforms existing training strategies. Our findings underscore the importance of carefully designed data construction in LALM development and offer practical insights for building robust, general-purpose LALMs.
- 19 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Audio, Speech and Language Processing
- Vol. 34, 2062-2076
Top Cited PapersDeSTA2.5-Audio: Toward General-Purpose Large Audio Language Model With Self-Generated Cross-Modal Alignment
We introduce DeSTA2.5-Audio, a general-purpose Large Audio Language Model (LALM) designed for robust auditory perception and instruction-following. Recent LALMs augment Large Language Models (LLMs) with auditory capabilities by training on large-scale audio-instruction datasets. However, existing LALMs have often suffered from the catastrophic forgetting of the LLM's original abilities. Therefore, balancing knowledge retention and audio perception has become a critical challenge. To address this, we revisit the data construction pipeline and propose a self-generated cross-modal alignment strategy in which the backbone LLM generates its own training targets, named DeSTA. This approach aims at preserving the LLM's native language proficiency thereby enabling zero-shot generalization without task-specific tuning. We construct DeSTA-AQA5M, a large-scale, task-agnostic dataset containing 5 million training samples derived from 7,000 hours of audio spanning 50 diverse datasets, including speech, environmental sounds, and music. DeSTA2.5-Audio achieves state-of-the-art or competitive performance across a wide range of audio-language benchmarks, including Dynamic-SUPERB, MMAU, SAKURA, Speech-IFEval, and VoiceBench. Comprehensive comparative studies demonstrate that our self-generated strategy outperforms existing training strategies. Our findings underscore the importance of carefully designed data construction in LALM development and offer practical insights for building robust, general-purpose LALMs.
- Open Access
- 26 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 7, 42-53
Contextual Attention for Robust Audio-Visual Emotion Recognition
Audio-visual emotion recognition (AVER) often performs well under ideal conditions but faces significant challenges in scenarios with missing modalities (e.g., missing frames of audio and/or video). Addressing these challenges is crucial for the effective deployment of AVER systems in human-computer interaction (HCI) applications, where robustness can significantly impact user experience. This study introduces a novel approach that enhances AVER robustness by leveraging a decoder-like summarizer structure. This structure processes audio and visual content and generates contextual summaries that effectively capture emotional cues even when modalities are degraded. To enhance system resilience against missing modalities, we integrate modality dropout during training, enabling the summarizer to adaptively handle these scenarios. We define the context summary length as the number of learnable query tokens used in the summarizer, a fixed hyperparameter in our model. We analyze how varying context summary lengths affect performance, identifying an optimal balance between compression and expressiveness. In addition to improving robustness, we systematically evaluate model calibration across emotions in current state-of-the-art (SOTA) AVER methods. Our experiments on the MSP-IMPROV and CREMA-D databases demonstrate that our model achieves superior performance across macro-, micro-, and weighted-F1 scores, both under ideal conditions and in scenarios with modality losses. Additionally, we conduct ablation studies to assess the impact of different context lengths on our summarizer structure in terms of overall AVER performance.
- 26 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 7, 42-53
Open AccessContextual Attention for Robust Audio-Visual Emotion Recognition
Audio-visual emotion recognition (AVER) often performs well under ideal conditions but faces significant challenges in scenarios with missing modalities (e.g., missing frames of audio and/or video). Addressing these challenges is crucial for the effective deployment of AVER systems in human-computer interaction (HCI) applications, where robustness can significantly impact user experience. This study introduces a novel approach that enhances AVER robustness by leveraging a decoder-like summarizer structure. This structure processes audio and visual content and generates contextual summaries that effectively capture emotional cues even when modalities are degraded. To enhance system resilience against missing modalities, we integrate modality dropout during training, enabling the summarizer to adaptively handle these scenarios. We define the context summary length as the number of learnable query tokens used in the summarizer, a fixed hyperparameter in our model. We analyze how varying context summary lengths affect performance, identifying an optimal balance between compression and expressiveness. In addition to improving robustness, we systematically evaluate model calibration across emotions in current state-of-the-art (SOTA) AVER methods. Our experiments on the MSP-IMPROV and CREMA-D databases demonstrate that our model achieves superior performance across macro-, micro-, and weighted-F1 scores, both under ideal conditions and in scenarios with modality losses. Additionally, we conduct ablation studies to assess the impact of different context lengths on our summarizer structure in terms of overall AVER performance.
- Open Access
- 15 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 6, 165-174
Jointly Learning From Unimodal and Multimodal-Rated Labels in Audio-Visual Emotion Recognition
Audio-visual emotion recognition (AVER) has been an important research area in human-computer interaction (HCI). Traditionally, audio-visual emotional datasets and corresponding models derive their ground truths from annotations obtained by raters after watching the audio-visual stimuli. This conventional method, however, neglects the nuanced human perception of emotional states, which varies when annotations are made under different emotional stimuli conditions-whether through unimodal or multimodal stimuli. This study investigates the potential for enhanced AVER system performance by integrating diverse levels of annotation stimuli, reflective of varying perceptual evaluations. We propose a two-stage training method to train models with the labels elicited by audio-only, face-only, and audio-visual stimuli. Our approach utilizes different levels of annotation stimuli according to which modality is present within different layers of the model, effectively modeling annotation at the unimodal and multi-modal levels to capture the full scope of emotion perception across unimodal and multimodal contexts. We conduct the experiments and evaluate the models on the CREMA-D emotion database. The proposed methods achieved the best performances in macro-/weighted-F1 scores. Additionally, we measure the model calibration, performance bias, and fairness metrics considering the age, gender, and race of the AVER systems.
- 15 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 6, 165-174
Open AccessJointly Learning From Unimodal and Multimodal-Rated Labels in Audio-Visual Emotion Recognition
Audio-visual emotion recognition (AVER) has been an important research area in human-computer interaction (HCI). Traditionally, audio-visual emotional datasets and corresponding models derive their ground truths from annotations obtained by raters after watching the audio-visual stimuli. This conventional method, however, neglects the nuanced human perception of emotional states, which varies when annotations are made under different emotional stimuli conditions-whether through unimodal or multimodal stimuli. This study investigates the potential for enhanced AVER system performance by integrating diverse levels of annotation stimuli, reflective of varying perceptual evaluations. We propose a two-stage training method to train models with the labels elicited by audio-only, face-only, and audio-visual stimuli. Our approach utilizes different levels of annotation stimuli according to which modality is present within different layers of the model, effectively modeling annotation at the unimodal and multi-modal levels to capture the full scope of emotion perception across unimodal and multimodal contexts. We conduct the experiments and evaluate the models on the CREMA-D emotion database. The proposed methods achieved the best performances in macro-/weighted-F1 scores. Additionally, we measure the model calibration, performance bias, and fairness metrics considering the age, gender, and race of the AVER systems.
- Open Access
- 7 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. 16 (1) , 41-55
Minority Views Matter: Evaluating Speech Emotion Classifiers With Human Subjective Annotations by an All-Inclusive Aggregation Rule
When selecting test data for subjective tasks, most studies define ground truth labels using aggregation methods such as the majority or plurality rules. These methods discard data points without consensus, making the test set easier than practical tasks where a prediction is needed for each sample. However, the discarded data points often express ambiguous cues that elicit coexisting traits perceived by annotators. This paper addresses the importance of considering all the annotations and samples in the data, highlighting that only showing the model's performance on an incomplete test set selected by using the majority or plurality rules can lead to bias in the models' performances. We focus on speech-emotion recognition (SER) tasks. We observe that traditional aggregation rules have a data loss ratio ranging from 5.63% to 89.17%. From this observation, we propose a flexible method named the all-inclusive aggregation rule to evaluate SER systems on the complete test data. We contrast traditional single-label formulations with a multi-label formulation to consider the coexistence of emotions. We show that training an SER model with the data selected by the all-inclusive aggregation rule shows consistently higher macro-F1 scores when tested in the entire test set, including ambiguous samples without agreement.
- 7 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. 16 (1) , 41-55
Open AccessMinority Views Matter: Evaluating Speech Emotion Classifiers With Human Subjective Annotations by an All-Inclusive Aggregation Rule
When selecting test data for subjective tasks, most studies define ground truth labels using aggregation methods such as the majority or plurality rules. These methods discard data points without consensus, making the test set easier than practical tasks where a prediction is needed for each sample. However, the discarded data points often express ambiguous cues that elicit coexisting traits perceived by annotators. This paper addresses the importance of considering all the annotations and samples in the data, highlighting that only showing the model's performance on an incomplete test set selected by using the majority or plurality rules can lead to bias in the models' performances. We focus on speech-emotion recognition (SER) tasks. We observe that traditional aggregation rules have a data loss ratio ranging from 5.63% to 89.17%. From this observation, we propose a flexible method named the all-inclusive aggregation rule to evaluate SER systems on the complete test data. We contrast traditional single-label formulations with a multi-label formulation to consider the coexistence of emotions. We show that training an SER model with the data selected by the all-inclusive aggregation rule shows consistently higher macro-F1 scores when tested in the entire test set, including ambiguous samples without agreement.