Emotional dialogue speech synthesis (EDSS) aims to generate expressive speech by leveraging the dialogue context between interlocutors. This is typically done by concatenating global representations of previous utterances as conditions for text-to-speech (TTS) systems. However, such approaches overlook the importance of integrating localized acoustic cues that convey emotion. To address this, we introduce a novel approach that utilizes a large language model (LLM) to generate holistic emotion tags based on prior dialogue context, while also pinpointing key words in the target utterance that align with the predicted emotional state. Furthermore, we enhance the emotional richness of synthesized speech by incorporating concentrated acoustic features of these key words through a novel selective audio masking loss function. This methodology not only improves emotional expressiveness, but also facilitates automatic emotion speech generation during inference by eliminating the need for manual emotion tag selection. Comprehensive subjective and objective evaluations and analyses demonstrate the effectiveness of the proposed approach.

Task-Oriented Dialogue (TOD) systems are designed to fulfill user requests through natural language interactions, yet existing systems often produce generic, monotonic responses that lack individuality and fail to adapt to users’ personal attributes. To address this, we introduce PicPersona-TOD, a novel dataset that incorporates user images as part of the persona, enabling personalized responses tailored to user-specific factors such as age or emotional context. This is facilitated by first impressions, dialogue policy-guided prompting, and the use of external knowledge to reduce hallucinations. Human evaluations confirm that our dataset enhances user experience, with personalized responses contributing to a more engaging interaction. Additionally, we introduce a new NLG model, Pictor, which not only personalizes responses, but also demonstrates robust performance across unseen domains.

This paper explores the task of language-agnostic speaker replication, a novel endeavor that seeks to replicate a speaker’s voice irrespective of the language they are speaking. Towards this end, we introduce a multi-level attention aggregation approach that systematically probes and amplifies various speaker-specific attributes in a hierarchical manner. Through rigorous evaluations across a wide range of scenarios including seen and unseen speakers conversing in seen and unseen lingua, we establish that our proposed model is able to achieve substantial speaker similarity, and is able to generalize to out-of-domain (OOD) cases.

The difficulty of acquiring abundant, high-quality data, especially in multi-lingual contexts, has sparked interest in addressing low-resource scenarios. Moreover, current literature rely on fixed expressions from language IDs, which results in the inadequate learning of language representations, and the failure to generate speech in unseen languages. To address these challenges, we propose a novel method that directly extracts linguistic features from audio input while effectively filtering out miscellaneous acoustic information including speaker-specific attributes like timbre. Subjective and objective evaluations affirm the effectiveness of our approach for multi-lingual text-to-speech, and highlight its superiority in low-resource transfer learning for previously unseen language.

Contemporary neural speech synthesis models have indeed demonstrated remarkable proficiency in synthetic speech generation as they have attained a level of quality comparable to that of human-produced speech. Nevertheless, it is important to note that these achievements have predominantly been verified within the context of high-resource languages such as English. Furthermore, the Tacotron and FastSpeech variants show substantial pausing errors when applied to the Korean language, which affects speech perception and naturalness. In order to address the aforementioned issues, we propose a novel framework that incorporates comprehensive modeling of both syntactic and acoustic cues that are associated with pausing patterns. Remarkably, our framework possesses the capability to consistently generate natural speech even for considerably more extended and intricate out-of-domain (OOD) sentences, despite its training on short audio clips. Architectural design choices are validated through comparisons with baseline models and ablation studies using subjective and objective metrics, thus confirming model performance.

Research on hate speech has predominantly revolved around the detection and interpretation from textual inputs, leaving verbal content largely unexplored. Moreover, while there has been some limited exploration into hate speech detection within verbal acoustic speech inputs, the aspect of interpretability has been overlooked. As such, we introduce a new task within the audio hate speech detection task domain - we specifically aim to identify specific time frames of hate speech within audio utterances. Towards this, we propose two different approaches, cascading and End-to-End (E2E). The first cascading approach initially converts audio to transcripts, identifies hate speech within these transcripts, and subsequently locates the corresponding audio time frames. Conversely, the second E2E approach processes audio utterances directly, which allows it to pinpoint hate speech within specific time frames. Moreover, due to the lack of explainable audio hate speech datasets that include frame-level rationales, we curated a synthetic audio dataset to train our models. We further validate these models on actual human speech utterances and we find that the E2E approach outperforms the cascading method in terms of audio frame Intersection over Union (IoU) metric. Furthermore, we observe that the inclusion of frame-level rationales significantly enhances hate speech detection accuracy for both E2E and cascading approaches.