Target-Speaker Voice Activity Detection (TS-VAD) utilizes a set of speaker profiles alongside an input audio signal to perform speaker diarization. While its superiority over conventional methods has been demonstrated, the method can suffer from errors in speaker profiles, as those profiles are typically obtained by running a traditional clustering-based diarization method over the input signal. This paper proposes an extension to TS-VAD, called Profile-Error-Tolerant TS- VAD (PETTSVAD), which is robust to such speaker profile errors. This is achieved by employing transformer-based TS-VAD that can handle a variable number of speakers and further introducing a set of additional pseudo-speaker profiles to handle speakers undetected during the first pass diarization. During training, we use speaker profiles estimated by multiple different clustering algorithms to reduce the mismatch between the training and testing conditions regarding speaker profiles. Experimental results show that PET-TSVAD consistently outperforms the existing TS-VAD method on both the VoxConverse and DIHARD-I datasets.

There is a rising interest and trend in research towards directly translating speech from one language to another, known as end-to-end speech-to-speech translation. However, most end-to-end models struggle to outperform cascade models, i.e., a pipeline framework by concatenating speech recognition, machine translation and text-to-speech models. The primary challenges stem from the inherent complexities involved in direct translation tasks and the scarcity of data. In this study, we introduce a novel model framework TransVIP that leverages diverse datasets in a cascade fashion yet facilitates end-to-end inference through joint probability. Furthermore, we propose two separated encoders to preserve the speaker’s voice characteristics and isochrony from the source speech during the translation process, making it highly suitable for scenarios such as video dubbing. Our experiments on the French-English language pair demonstrate that our model outperforms the current state-of-the-art speech-to-speech translation model.

Recent advancements in zero-shot text-to-speech (TTS) modeling have led to significant strides in generating high-fidelity and diverse speech. However, dialogue generation, along with achieving human-like naturalness in speech, continues to be a challenge. In this paper, we introduce CoVoMix: Conversational Voice Mixture Generation, a novel model for zero-shot, human-like, multi-speaker, multi-round dialogue speech generation. CoVoMix first converts dialogue text into multiple streams of discrete tokens, with each token stream representing semantic information for individual talkers. These token streams are then fed into a flow-matching based acoustic model to generate mixed mel-spectrograms. Finally, the speech waveforms are produced using a HiFi-GAN model. Furthermore, we devise a comprehensive set of metrics for measuring the effectiveness of dialogue modeling and generation. Our experimental results show that CoVoMix can generate dialogues that are not only human-like in their naturalness and coherence but also involve multiple talkers engaging in multiple rounds of conversation. This is exemplified by instances generated in a single channel where one speaker's utterance is seamlessly mixed with another's interjections or laughter, indicating the latter's role as an attentive listener. Audio samples are enclosed in the supplementary.

Neural audio codec tokens serve as the fundamental building blocks for speech language model (SLM)-based speech generation. However, there is no systematic understanding on how the codec system affects the speech generation performance of the SLM. In this work, we examine codec tokens within SLM framework for speech generation to provide insights for effective codec design. We retrain existing high-performing neural codec models on the same data set and loss functions to compare their performance in a uniform setting. We integrate codec tokens into two SLM systems: masked-based parallel speech generation system and an auto-regressive (AR) plus non-auto-regressive (NAR) model-based system. Our findings indicate that better speech reconstruction in codec systems does not guarantee improved speech generation in SLM. A high-quality codec decoder is crucial for natural speech production in SLM, while speech intelligibility depends more on quantization mechanism.

This study addresses the problem of single-channel Automatic Speech Recognition of a target speaker within an overlap speech scenario. In the proposed method, the hidden representations in the acoustic model are modulated by speaker auxiliary information to recognize only the desired speaker. Affine transformation layers are inserted into the acoustic model network to integrate speaker information with the acoustic features. The speaker conditioning process allows the acoustic model to perform computation in the context of target-speaker auxiliary information. The proposed speaker conditioning method is a general approach and can be applied to any acoustic model architecture. Here, we employ speaker conditioning on a ResNet acoustic model. Experiments on the WSJ corpus show that the proposed speaker conditioning method is an effective solution to fuse speaker auxiliary information with acoustic features for multi-speaker speech recognition, achieving +9% and +20% relative WER reduction for clean and overlap speech scenarios, respectively, compared to the original ResNet acoustic model baseline.

Most current speech technology systems are designed to operate well even in the presence of multiple active speakers. However, most solutions assume that the number of co-current speakers is known. Unfortunately, this information might not always be available in real-world applications. In this study, we propose a real-time, single-channel attention-guided Convolutional Neural Network (CNN) to estimate the number of active speakers in overlapping speech. The proposed system extracts higher-level information from the speech spectral content using a CNN model. Next, the attention mechanism summarizes the extracted information into a compact feature vector without losing critical information. Finally, the active speakers are classified using a fully connected network. Experiments on simulated overlapping speech using WSJ corpus show that the attention solution is shown to improve the performance by almost 3% absolute over conventional temporal average pooling. The proposed Attention-guided CNN achieves 76.15% for both Weighted Accuracy and average Recall, and 75.80% Precision on speech segments as short as 20 frames (i.e., 200 ms). All the classification metrics exceed 92% for the attention-guided model in offline scenarios where the input signal is more than 100 frames long (i.e., 1s).

Speech technology systems such as Automatic Speech Recognition (ASR), speaker diarization, speaker recognition, and speech synthesis have advanced significantly by the emergence of deep learning techniques. However, none of these voice-enabled systems perform well in natural environmental circumstances, specifically in situations where one or more potential interfering talkers are involved. Therefore, overlapping speech detection has become an important front-end triage step for speech technology applications. This is crucial for large-scale datasets where manual labeling in not possible. A block-based CNN architecture is proposed to address modeling overlapping speech in audio streams with frames as short as 25 ms. The proposed architecture is robust to both: (i) shifts in distribution of network activations due to the change in network parameters during training, (ii) local variations from the input features caused by feature extraction, environmental noise, or room interference. We also investigate the effect of alternate input features including spectral magnitude, MFCC, MFB, and pyknogram on both computational time and classification performance. Evaluation is performed on simulated overlapping speech signals based on the GRID corpus. The experimental results highlight the capability of the proposed system in detecting overlapping speech frames with 90.5% accuracy, 93.5% precision, 92.7% recall, and 92.8% Fscore on same gender overlapped speech. For opposite gender cases, the network scores exceed 95% in all the classification metrics.

Naturalistic speech recordings usually contain speech signals from multiple speakers. This phenomenon can degrade the performance of speech technologies due to the complexity of tracing and recognizing individual speakers. In this study, we investigate the detection of overlapping speech on segments as short as 25 ms using Convolutional Neural Networks. We evaluate the detection performance using different spectral features, and show that pyknogram features outperforms other commonly used speech features. The proposed system can predict overlapping speech with an accuracy of 84% and Fs-core of 88% on a dataset of mixed speech generated based on the GRID dataset.

Single-microphone, speaker-independent speech separation is normally performed through two steps: (i) separating the specific speech sources, and (ii) determining the best output-label assignment to find the separation error. The second step is the main obstacle in training neural networks for speech separation. Recently proposed Permutation Invariant Training (PIT) addresses this problem by determining the output-label assignment which minimizes the separation error. In this study, we show that a major drawback of this technique is the overconfident choice of the output-label assignment, especially in the initial steps of training when the network generates unreliable outputs. To solve this problem, we propose Probabilistic PIT (Prob-PIT) which considers the output-label permutation as a discrete latent random variable with a uniform prior distribution. Prob-PIT defines a log-likelihood function based on the prior distributions and the separation errors of all permutations; it trains the speech separation networks by maximizing the log-likelihood function. Prob-PIT can be easily implemented by replacing the minimum function of PIT with a soft-minimum function. We evaluate our approach for speech separation on both TIMIT and CHiME datasets. The results show that the proposed method significantly outperforms PIT in terms of Signal to Distortion Ratio and Signal to Interference Ratio.