Recent advancements in reference-free learned metrics for open-domain dialogue evaluation have been driven by the progress in pre-trained language models and the availability of dialogue data with high-quality human annotations. However, current studies predominantly concentrate on English dialogues, and the generalization of these metrics to other languages has not been fully examined. This is largely due to the absence of a multilingual dialogue evaluation benchmark. To address the issue, we introduce xDial-Eval, built on top of open-source English dialogue evaluation datasets. xDial-Eval includes 12 turn-level and 6 dialogue-level English datasets, comprising 14930 annotated turns and 8691 annotated dialogues respectively. The English dialogue data are extended to nine other languages with commercial machine translation systems. On xDial-Eval, we conduct comprehensive analyses of previous BERT-based metrics and the recently-emerged large language models. Lastly, we establish strong self-supervised and multilingual baselines. In terms of average Pearson correlations over all datasets and languages, the best baseline outperforms OpenAI's ChatGPT by absolute improvements of 6.5% and 4.6% at the turn and dialogue levels respectively, albeit with much fewer parameters. The data and code are publicly available at https://github.com/e0397123/xDial-Eval.
For speaker recognition, it is difficult to extract an accurate speaker representation from speech because of its mixture of speaker traits and content. This paper proposes a disentanglement framework that simultaneously models speaker traits and content variability in speech. It is realized with the use of three Gaussian inference layers, each consisting of a learnable transition model that extracts distinct speech components. Notably, a strengthened transition model is specifically designed to model complex speech dynamics. We also propose a self-supervision method to dynamically disentangle content without the use of labels other than speaker identities. The efficacy of the proposed framework is validated via experiments conducted on the VoxCeleb and SITW datasets with 9.56% and 8.24% average reductions in EER and minDCF, respectively. Since neither additional model training nor data is specifically needed, it is easily applicable in practical use.
Current speaker recognition systems primarily rely on supervised approaches, constrained by the scale of labeled datasets. To boost the system performance, researchers leverage large pretrained models such as WavLM to transfer learned high-level features to the downstream speaker recognition task. However, this approach introduces extra parameters as the pretrained model remains in the inference stage. Another group of researchers directly apply self-supervised methods such as DINO to speaker embedding learning, yet they have not explored its potential on large-scale in-the-wild datasets. In this paper, we present the effectiveness of DINO training on the large-scale WenetSpeech dataset and its transferability in enhancing the supervised system performance on the CNCeleb dataset. Additionally, we introduce a confidence-based data filtering algorithm to remove unreliable data from the pretraining dataset, leading to better performance with less training data. The associated pretrained models, confidence files, pretraining and finetuning scripts will be made available in the Wespeaker toolkit.
This paper explores the imperative need and methodology for developing a localized Large Language Model (LLM) tailored for Arabic, a language with unique cultural characteristics that are not adequately addressed by current mainstream models like ChatGPT. Key concerns additionally arise when considering cultural sensitivity and local values. To this end, the paper outlines a packaged solution, including further pre-training with Arabic texts, supervised fine-tuning (SFT) using native Arabic instructions and GPT-4 responses in Arabic, and reinforcement learning with AI feedback (RLAIF) using a reward model that is sensitive to local culture and values. The objective is to train culturally aware and value-aligned Arabic LLMs that can serve the diverse application-specific needs of Arabic-speaking communities. Extensive evaluations demonstrated that the resulting LLM called `AceGPT' is the SOTA open Arabic LLM in various benchmarks, including instruction-following benchmark (i.e., Arabic Vicuna-80 and Arabic AlpacaEval), knowledge benchmark (i.e., Arabic MMLU and EXAMs), as well as the newly-proposed Arabic cultural \& value alignment benchmark. Notably, AceGPT outperforms ChatGPT in the popular Vicuna-80 benchmark when evaluated with GPT-4, despite the benchmark's limited scale. % Natural Language Understanding (NLU) benchmark (i.e., ALUE) Codes, data, and models are in https://github.com/FreedomIntelligence/AceGPT.
Text-based speech editing (TSE) techniques are designed to enable users to edit the output audio by modifying the input text transcript instead of the audio itself. Despite much progress in neural network-based TSE techniques, the current techniques have focused on reducing the difference between the generated speech segment and the reference target in the editing region, ignoring its local and global fluency in the context and original utterance. To maintain the speech fluency, we propose a fluency speech editing model, termed \textit{FluentEditor}, by considering fluency-aware training criterion in the TSE training. Specifically, the \textit{acoustic consistency constraint} aims to smooth the transition between the edited region and its neighboring acoustic segments consistent with the ground truth, while the \textit{prosody consistency constraint} seeks to ensure that the prosody attributes within the edited regions remain consistent with the overall style of the original utterance. The subjective and objective experimental results on VCTK demonstrate that our \textit{FluentEditor} outperforms all advanced baselines in terms of naturalness and fluency. The audio samples and code are available at \url{https://github.com/Ai-S2-Lab/FluentEditor}.
Prosodic phrasing is crucial to the naturalness and intelligibility of end-to-end Text-to-Speech (TTS). There exist both linguistic and emotional prosody in natural speech. As the study of prosodic phrasing has been linguistically motivated, prosodic phrasing for expressive emotion rendering has not been well studied. In this paper, we propose an emotion-aware prosodic phrasing model, termed \textit{EmoPP}, to mine the emotional cues of utterance accurately and predict appropriate phrase breaks. We first conduct objective observations on the ESD dataset to validate the strong correlation between emotion and prosodic phrasing. Then the objective and subjective evaluations show that the EmoPP outperforms all baselines and achieves remarkable performance in terms of emotion expressiveness. The audio samples and the code are available at \url{https://github.com/AI-S2-Lab/EmoPP}.
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/.
Brain-inspired spiking neural networks (SNNs) have demonstrated great potential for temporal signal processing. However, their performance in speech processing remains limited due to the lack of an effective auditory front-end. To address this limitation, we introduce Spiking-LEAF, a learnable auditory front-end meticulously designed for SNN-based speech processing. Spiking-LEAF combines a learnable filter bank with a novel two-compartment spiking neuron model called IHC-LIF. The IHC-LIF neurons draw inspiration from the structure of inner hair cells (IHC) and they leverage segregated dendritic and somatic compartments to effectively capture multi-scale temporal dynamics of speech signals. Additionally, the IHC-LIF neurons incorporate the lateral feedback mechanism along with spike regularization loss to enhance spike encoding efficiency. On keyword spotting and speaker identification tasks, the proposed Spiking-LEAF outperforms both SOTA spiking auditory front-ends and conventional real-valued acoustic features in terms of classification accuracy, noise robustness, and encoding efficiency.
Target speaker extraction aims to extract the speech of a specific speaker from a multi-talker mixture as specified by an auxiliary reference. Most studies focus on the scenario where the target speech is highly overlapped with the interfering speech. However, this scenario only accounts for a small percentage of real-world conversations. In this paper, we aim at the sparsely overlapped scenarios in which the auxiliary reference needs to perform two tasks simultaneously: detect the activity of the target speaker and disentangle the active speech from any interfering speech. We propose an audio-visual speaker extraction model named ActiveExtract, which leverages speaking activity from audio-visual active speaker detection (ASD). The ASD directly provides the frame-level activity of the target speaker, while its intermediate feature representation is trained to discriminate speech-lip synchronization that could be used for speaker disentanglement. Experimental results show our model outperforms baselines across various overlapping ratios, achieving an average improvement of more than 4 dB in terms of SI-SNR.