Daily Papers

Creating universal speaker encoders which are robust for different acoustic and speech duration conditions is a big challenge today. According to our observations systems trained on short speech segments are optimal for short phrase speaker verification and systems trained on long segments are superior for long segments verification. A system trained simultaneously on pooled short and long speech segments does not give optimal verification results and usually degrades both for short and long segments. This paper addresses the problem of creating universal speaker encoders for different speech segments duration. We describe our simple recipe for training universal speaker encoder for any type of selected neural network architecture. According to our evaluation results of wav2vec-TDNN based systems obtained for NIST SRE and VoxCeleb1 benchmarks the proposed universal encoder provides speaker verification improvements in case of different enrollment and test speech segment duration. The key feature of the proposed encoder is that it has the same inference time as the selected neural network architecture.

We introduce MiniMax-Speech, an autoregressive Transformer-based Text-to-Speech (TTS) model that generates high-quality speech. A key innovation is our learnable speaker encoder, which extracts timbre features from a reference audio without requiring its transcription. This enables MiniMax-Speech to produce highly expressive speech with timbre consistent with the reference in a zero-shot manner, while also supporting one-shot voice cloning with exceptionally high similarity to the reference voice. In addition, the overall quality of the synthesized audio is enhanced through the proposed Flow-VAE. Our model supports 32 languages and demonstrates excellent performance across multiple objective and subjective evaluations metrics. Notably, it achieves state-of-the-art (SOTA) results on objective voice cloning metrics (Word Error Rate and Speaker Similarity) and has secured the top position on the public TTS Arena leaderboard. Another key strength of MiniMax-Speech, granted by the robust and disentangled representations from the speaker encoder, is its extensibility without modifying the base model, enabling various applications such as: arbitrary voice emotion control via LoRA; text to voice (T2V) by synthesizing timbre features directly from text description; and professional voice cloning (PVC) by fine-tuning timbre features with additional data. We encourage readers to visit https://minimax-ai.github.io/tts_tech_report for more examples.

We describe a neural network-based system for text-to-speech (TTS) synthesis that is able to generate speech audio in the voice of many different speakers, including those unseen during training. Our system consists of three independently trained components: (1) a speaker encoder network, trained on a speaker verification task using an independent dataset of noisy speech from thousands of speakers without transcripts, to generate a fixed-dimensional embedding vector from seconds of reference speech from a target speaker; (2) a sequence-to-sequence synthesis network based on Tacotron 2, which generates a mel spectrogram from text, conditioned on the speaker embedding; (3) an auto-regressive WaveNet-based vocoder that converts the mel spectrogram into a sequence of time domain waveform samples. We demonstrate that the proposed model is able to transfer the knowledge of speaker variability learned by the discriminatively-trained speaker encoder to the new task, and is able to synthesize natural speech from speakers that were not seen during training. We quantify the importance of training the speaker encoder on a large and diverse speaker set in order to obtain the best generalization performance. Finally, we show that randomly sampled speaker embeddings can be used to synthesize speech in the voice of novel speakers dissimilar from those used in training, indicating that the model has learned a high quality speaker representation.

Probabilistic linear discriminant analysis (PLDA) or cosine similarity have been widely used in traditional speaker verification systems as back-end techniques to measure pairwise similarities. To make better use of multiple enrollment utterances, we propose a novel attention back-end model, which can be used for both text-independent (TI) and text-dependent (TD) speaker verification, and employ scaled-dot self-attention and feed-forward self-attention networks as architectures that learn the intra-relationships of the enrollment utterances. In order to verify the proposed attention back-end, we conduct a series of experiments on CNCeleb and VoxCeleb datasets by combining it with several sate-of-the-art speaker encoders including TDNN and ResNet. Experimental results using multiple enrollment utterances on CNCeleb show that the proposed attention back-end model leads to lower EER and minDCF score than the PLDA and cosine similarity counterparts for each speaker encoder and an experiment on VoxCeleb indicate that our model can be used even for single enrollment case.

In this paper, we introduce a zero-shot Voice Transfer (VT) module that can be seamlessly integrated into a multi-lingual Text-to-speech (TTS) system to transfer an individual's voice across languages. Our proposed VT module comprises a speaker-encoder that processes reference speech, a bottleneck layer, and residual adapters, connected to preexisting TTS layers. We compare the performance of various configurations of these components and report Mean Opinion Score (MOS) and Speaker Similarity across languages. Using a single English reference speech per speaker, we achieve an average voice transfer similarity score of 73% across nine target languages. Vocal characteristics contribute significantly to the construction and perception of individual identity. The loss of one's voice, due to physical or neurological conditions, can lead to a profound sense of loss, impacting one's core identity. As a case study, we demonstrate that our approach can not only transfer typical speech but also restore the voices of individuals with dysarthria, even when only atypical speech samples are available - a valuable utility for those who have never had typical speech or banked their voice. Cross-lingual typical audio samples, plus videos demonstrating voice restoration for dysarthric speakers are available here (google.github.io/tacotron/publications/zero_shot_voice_transfer).

Speech synthesis technology has witnessed significant advancements in recent years, enabling the creation of natural and expressive synthetic speech. One area of particular interest is the generation of synthetic child speech, which presents unique challenges due to children's distinct vocal characteristics and developmental stages. This paper presents a novel approach that leverages the Fastpitch text-to-speech (TTS) model for generating high-quality synthetic child speech. This study uses the transfer learning training pipeline. The approach involved finetuning a multi-speaker TTS model to work with child speech. We use the cleaned version of the publicly available MyST dataset (55 hours) for our finetuning experiments. We also release a prototype dataset of synthetic speech samples generated from this research together with model code to support further research. By using a pretrained MOSNet, we conducted an objective assessment that showed a significant correlation between real and synthetic child voices. Additionally, to validate the intelligibility of the generated speech, we employed an automatic speech recognition (ASR) model to compare the word error rates (WER) of real and synthetic child voices. The speaker similarity between the real and generated speech is also measured using a pretrained speaker encoder.

This work presents FreeSVC, a promising multilingual singing voice conversion approach that leverages an enhanced VITS model with Speaker-invariant Clustering (SPIN) for better content representation and the State-of-the-Art (SOTA) speaker encoder ECAPA2. FreeSVC incorporates trainable language embeddings to handle multiple languages and employs an advanced speaker encoder to disentangle speaker characteristics from linguistic content. Designed for zero-shot learning, FreeSVC enables cross-lingual singing voice conversion without extensive language-specific training. We demonstrate that a multilingual content extractor is crucial for optimal cross-language conversion. Our source code and models are publicly available.

Better disentanglement of speech representation is essential to improve the quality of voice conversion. Recently contrastive learning is applied to voice conversion successfully based on speaker labels. However, the performance of model will reduce in conversion between similar speakers. Hence, we propose an augmented negative sample selection to address the issue. Specifically, we create hard negative samples based on the proposed speaker fusion module to improve learning ability of speaker encoder. Furthermore, considering the fine-grain modeling of speaker style, we employ a reference encoder to extract fine-grained style and conduct the augmented contrastive learning on global style. The experimental results show that the proposed method outperforms previous work in voice conversion tasks.

We propose Cotatron, a transcription-guided speech encoder for speaker-independent linguistic representation. Cotatron is based on the multispeaker TTS architecture and can be trained with conventional TTS datasets. We train a voice conversion system to reconstruct speech with Cotatron features, which is similar to the previous methods based on Phonetic Posteriorgram (PPG). By training and evaluating our system with 108 speakers from the VCTK dataset, we outperform the previous method in terms of both naturalness and speaker similarity. Our system can also convert speech from speakers that are unseen during training, and utilize ASR to automate the transcription with minimal reduction of the performance. Audio samples are available at https://mindslab-ai.github.io/cotatron, and the code with a pre-trained model will be made available soon.

Zero-shot text-to-speech (TTS) synthesis aims to clone any unseen speaker's voice without adaptation parameters. By quantizing speech waveform into discrete acoustic tokens and modeling these tokens with the language model, recent language model-based TTS models show zero-shot speaker adaptation capabilities with only a 3-second acoustic prompt of an unseen speaker. However, they are limited by the length of the acoustic prompt, which makes it difficult to clone personal speaking style. In this paper, we propose a novel zero-shot TTS model with the multi-scale acoustic prompts based on a neural codec language model VALL-E. A speaker-aware text encoder is proposed to learn the personal speaking style at the phoneme-level from the style prompt consisting of multiple sentences. Following that, a VALL-E based acoustic decoder is utilized to model the timbre from the timbre prompt at the frame-level and generate speech. The experimental results show that our proposed method outperforms baselines in terms of naturalness and speaker similarity, and can achieve better performance by scaling out to a longer style prompt.

End-to-end speaker diarization for an unknown number of speakers is addressed in this paper. Recently proposed end-to-end speaker diarization outperformed conventional clustering-based speaker diarization, but it has one drawback: it is less flexible in terms of the number of speakers. This paper proposes a method for encoder-decoder based attractor calculation (EDA), which first generates a flexible number of attractors from a speech embedding sequence. Then, the generated multiple attractors are multiplied by the speech embedding sequence to produce the same number of speaker activities. The speech embedding sequence is extracted using the conventional self-attentive end-to-end neural speaker diarization (SA-EEND) network. In a two-speaker condition, our method achieved a 2.69 % diarization error rate (DER) on simulated mixtures and a 8.07 % DER on the two-speaker subset of CALLHOME, while vanilla SA-EEND attained 4.56 % and 9.54 %, respectively. In unknown numbers of speakers conditions, our method attained a 15.29 % DER on CALLHOME, while the x-vector-based clustering method achieved a 19.43 % DER.

Recently, text-guided content generation has received extensive attention. In this work, we explore the possibility of text description-based speaker generation, i.e., using text prompts to control the speaker generation process. Specifically, we propose PromptSpeaker, a text-guided speaker generation system. PromptSpeaker consists of a prompt encoder, a zero-shot VITS, and a Glow model, where the prompt encoder predicts a prior distribution based on the text description and samples from this distribution to obtain a semantic representation. The Glow model subsequently converts the semantic representation into a speaker representation, and the zero-shot VITS finally synthesizes the speaker's voice based on the speaker representation. We verify that PromptSpeaker can generate speakers new from the training set by objective metrics, and the synthetic speaker voice has reasonable subjective matching quality with the speaker prompt.

We propose UnitSpeech, a speaker-adaptive speech synthesis method that fine-tunes a diffusion-based text-to-speech (TTS) model using minimal untranscribed data. To achieve this, we use the self-supervised unit representation as a pseudo transcript and integrate the unit encoder into the pre-trained TTS model. We train the unit encoder to provide speech content to the diffusion-based decoder and then fine-tune the decoder for speaker adaptation to the reference speaker using a single <unit, speech> pair. UnitSpeech performs speech synthesis tasks such as TTS and voice conversion (VC) in a personalized manner without requiring model re-training for each task. UnitSpeech achieves comparable and superior results on personalized TTS and any-to-any VC tasks compared to previous baselines. Our model also shows widespread adaptive performance on real-world data and other tasks that use a unit sequence as input.

In this paper, we propose SC-GlowTTS: an efficient zero-shot multi-speaker text-to-speech model that improves similarity for speakers unseen during training. We propose a speaker-conditional architecture that explores a flow-based decoder that works in a zero-shot scenario. As text encoders, we explore a dilated residual convolutional-based encoder, gated convolutional-based encoder, and transformer-based encoder. Additionally, we have shown that adjusting a GAN-based vocoder for the spectrograms predicted by the TTS model on the training dataset can significantly improve the similarity and speech quality for new speakers. Our model converges using only 11 speakers, reaching state-of-the-art results for similarity with new speakers, as well as high speech quality.

Speaker identification in multilingual settings presents unique challenges, particularly when conventional models are predominantly trained on English data. In this paper, we propose WSI (Whisper Speaker Identification), a framework that repurposes the encoder of the Whisper automatic speech recognition model pre trained on extensive multilingual data to generate robust speaker embeddings via a joint loss optimization strategy that leverages online hard triplet mining and self supervised Normalized Temperature-scaled Cross Entropy loss. By capitalizing on Whisper language-agnostic acoustic representations, our approach effectively distinguishes speakers across diverse languages and recording conditions. Extensive evaluations on multiple corpora, including VoxTube (multilingual), JVS (Japanese), CallHome (German, Spanish, Chinese, and Japanese), and Voxconverse (English), demonstrate that WSI consistently outperforms state-of-the-art baselines, namely Pyannote Embedding, ECAPA TDNN, and Xvector, in terms of lower equal error rates and higher AUC scores. These results validate our hypothesis that a multilingual pre-trained ASR encoder, combined with joint loss optimization, substantially improves speaker identification performance in non-English languages.

This work presents a framework based on feature disentanglement to learn speaker embeddings that are robust to environmental variations. Our framework utilises an auto-encoder as a disentangler, dividing the input speaker embedding into components related to the speaker and other residual information. We employ a group of objective functions to ensure that the auto-encoder's code representation - used as the refined embedding - condenses only the speaker characteristics. We show the versatility of our framework through its compatibility with any existing speaker embedding extractor, requiring no structural modifications or adaptations for integration. We validate the effectiveness of our framework by incorporating it into two popularly used embedding extractors and conducting experiments across various benchmarks. The results show a performance improvement of up to 16%. We release our code for this work to be available https://github.com/kaistmm/voxceleb-disentangler

In this paper, we proposed Adapitch, a multi-speaker TTS method that makes adaptation of the supervised module with untranscribed data. We design two self supervised modules to train the text encoder and mel decoder separately with untranscribed data to enhance the representation of text and mel. To better handle the prosody information in a synthesized voice, a supervised TTS module is designed conditioned on content disentangling of pitch, text, and speaker. The training phase was separated into two parts, pretrained and fixed the text encoder and mel decoder with unsupervised mode, then the supervised mode on the disentanglement of TTS. Experiment results show that the Adaptich achieved much better quality than baseline methods.

Primary goal of an emotional voice conversion (EVC) system is to convert the emotion of a given speech signal from one style to another style without modifying the linguistic content of the signal. Most of the state-of-the-art approaches convert emotions for seen speaker-emotion combinations only. In this paper, we tackle the problem of converting the emotion of speakers whose only neutral data are present during the time of training and testing (i.e., unseen speaker-emotion combinations). To this end, we extend a recently proposed StartGANv2-VC architecture by utilizing dual encoders for learning the speaker and emotion style embeddings separately along with dual domain source classifiers. For achieving the conversion to unseen speaker-emotion combinations, we propose a Virtual Domain Pairing (VDP) training strategy, which virtually incorporates the speaker-emotion pairs that are not present in the real data without compromising the min-max game of a discriminator and generator in adversarial training. We evaluate the proposed method using a Hindi emotional database.

A general disentanglement-based speaker anonymization system typically separates speech into content, speaker, and prosody features using individual encoders. This paper explores how to adapt such a system when a new speech attribute, for example, emotion, needs to be preserved to a greater extent. While existing systems are good at anonymizing speaker embeddings, they are not designed to preserve emotion. Two strategies for this are examined. First, we show that integrating emotion embeddings from a pre-trained emotion encoder can help preserve emotional cues, even though this approach slightly compromises privacy protection. Alternatively, we propose an emotion compensation strategy as a post-processing step applied to anonymized speaker embeddings. This conceals the original speaker's identity and reintroduces the emotional traits lost during speaker embedding anonymization. Specifically, we model the emotion attribute using support vector machines to learn separate boundaries for each emotion. During inference, the original speaker embedding is processed in two ways: one, by an emotion indicator to predict emotion and select the emotion-matched SVM accurately; and two, by a speaker anonymizer to conceal speaker characteristics. The anonymized speaker embedding is then modified along the corresponding SVM boundary towards an enhanced emotional direction to save the emotional cues. The proposed strategies are also expected to be useful for adapting a general disentanglement-based speaker anonymization system to preserve other target paralinguistic attributes, with potential for a range of downstream tasks.

We propose Sortformer, a novel neural model for speaker diarization, trained with unconventional objectives compared to existing end-to-end diarization models. The permutation problem in speaker diarization has long been regarded as a critical challenge. Most prior end-to-end diarization systems employ permutation invariant loss (PIL), which optimizes for the permutation that yields the lowest error. In contrast, we introduce Sort Loss, which enables a diarization model to autonomously resolve permutation, with or without PIL. We demonstrate that combining Sort Loss and PIL achieves performance competitive with state-of-the-art end-to-end diarization models trained exclusively with PIL. Crucially, we present a streamlined multispeaker ASR architecture that leverages Sortformer as a speaker supervision model, embedding speaker label estimation within the ASR encoder state using a sinusoidal kernel function. This approach resolves the speaker permutation problem through sorted objectives, effectively bridging speaker-label timestamps and speaker tokens. In our experiments, we show that the proposed multispeaker ASR architecture, enhanced with speaker supervision, improves performance via adapter techniques. Code and trained models will be made publicly available via the NVIDIA NeMo framework

Automatic lip-reading (ALR) aims to automatically transcribe spoken content from a speaker's silent lip motion captured in video. Current mainstream lip-reading approaches only use a single visual encoder to model input videos of a single scale. In this paper, we propose to enhance lip-reading by incorporating multi-scale video data and multi-encoder. Specifically, we first propose a novel multi-scale lip motion extraction algorithm based on the size of the speaker's face and an Enhanced ResNet3D visual front-end (VFE) to extract lip features at different scales. For the multi-encoder, in addition to the mainstream Transformer and Conformer, we also incorporate the recently proposed Branchformer and E-Branchformer as visual encoders. In the experiments, we explore the influence of different video data scales and encoders on ALR system performance and fuse the texts transcribed by all ALR systems using recognizer output voting error reduction (ROVER). Finally, our proposed approach placed second in the ICME 2024 ChatCLR Challenge Task 2, with a 21.52% reduction in character error rate (CER) compared to the official baseline on the evaluation set.

Speaker adaptation in text-to-speech synthesis (TTS) is to finetune a pre-trained TTS model to adapt to new target speakers with limited data. While much effort has been conducted towards this task, seldom work has been performed for low computational resource scenarios due to the challenges raised by the requirement of the lightweight model and less computational complexity. In this paper, a tiny VITS-based TTS model, named AdaVITS, for low computing resource speaker adaptation is proposed. To effectively reduce parameters and computational complexity of VITS, an iSTFT-based wave construction decoder is proposed to replace the upsampling-based decoder which is resource-consuming in the original VITS. Besides, NanoFlow is introduced to share the density estimate across flow blocks to reduce the parameters of the prior encoder. Furthermore, to reduce the computational complexity of the textual encoder, scaled-dot attention is replaced with linear attention. To deal with the instability caused by the simplified model, instead of using the original text encoder, phonetic posteriorgram (PPG) is utilized as linguistic feature via a text-to-PPG module, which is then used as input for the encoder. Experiment shows that AdaVITS can generate stable and natural speech in speaker adaptation with 8.97M model parameters and 0.72GFlops computational complexity.

Motivated by the success of T5 (Text-To-Text Transfer Transformer) in pre-trained natural language processing models, we propose a unified-modal SpeechT5 framework that explores the encoder-decoder pre-training for self-supervised speech/text representation learning. The SpeechT5 framework consists of a shared encoder-decoder network and six modal-specific (speech/text) pre/post-nets. After preprocessing the input speech/text through the pre-nets, the shared encoder-decoder network models the sequence-to-sequence transformation, and then the post-nets generate the output in the speech/text modality based on the output of the decoder. Leveraging large-scale unlabeled speech and text data, we pre-train SpeechT5 to learn a unified-modal representation, hoping to improve the modeling capability for both speech and text. To align the textual and speech information into this unified semantic space, we propose a cross-modal vector quantization approach that randomly mixes up speech/text states with latent units as the interface between encoder and decoder. Extensive evaluations show the superiority of the proposed SpeechT5 framework on a wide variety of spoken language processing tasks, including automatic speech recognition, speech synthesis, speech translation, voice conversion, speech enhancement, and speaker identification. We release our code and model at https://github.com/microsoft/SpeechT5.

Multilingual automatic speech recognition (ASR) in the medical domain serves as a foundational task for various downstream applications such as speech translation, spoken language understanding, and voice-activated assistants. This technology enhances patient care by enabling efficient communication across language barriers, alleviating specialized workforce shortages, and facilitating improved diagnosis and treatment, particularly during pandemics. In this work, we introduce MultiMed, a collection of small-to-large end-to-end ASR models for the medical domain, spanning five languages: Vietnamese, English, German, French, and Mandarin Chinese, together with the corresponding real-world ASR dataset. To our best knowledge, MultiMed stands as the largest and the first multilingual medical ASR dataset, in terms of total duration, number of speakers, diversity of diseases, recording conditions, speaker roles, unique medical terms, accents, and ICD-10 codes. Secondly, we establish the empirical baselines, present the first reproducible study of multilinguality in medical ASR, conduct a layer-wise ablation study for end-to-end ASR training, and provide the first linguistic analysis for multilingual medical ASR. All code, data, and models are available online https://github.com/leduckhai/MultiMed/tree/master/MultiMed

Generating high-quality and person-generic visual dubbing remains a challenge. Recent innovation has seen the advent of a two-stage paradigm, decoupling the rendering and lip synchronization process facilitated by intermediate representation as a conduit. Still, previous methodologies rely on rough landmarks or are confined to a single speaker, thus limiting their performance. In this paper, we propose DiffDub: Diffusion-based dubbing. We first craft the Diffusion auto-encoder by an inpainting renderer incorporating a mask to delineate editable zones and unaltered regions. This allows for seamless filling of the lower-face region while preserving the remaining parts. Throughout our experiments, we encountered several challenges. Primarily, the semantic encoder lacks robustness, constricting its ability to capture high-level features. Besides, the modeling ignored facial positioning, causing mouth or nose jitters across frames. To tackle these issues, we employ versatile strategies, including data augmentation and supplementary eye guidance. Moreover, we encapsulated a conformer-based reference encoder and motion generator fortified by a cross-attention mechanism. This enables our model to learn person-specific textures with varying references and reduces reliance on paired audio-visual data. Our rigorous experiments comprehensively highlight that our ground-breaking approach outpaces existing methods with considerable margins and delivers seamless, intelligible videos in person-generic and multilingual scenarios.

We introduce a self-supervised speech pre-training method called TERA, which stands for Transformer Encoder Representations from Alteration. Recent approaches often learn by using a single auxiliary task like contrastive prediction, autoregressive prediction, or masked reconstruction. Unlike previous methods, we use alteration along three orthogonal axes to pre-train Transformer Encoders on a large amount of unlabeled speech. The model learns through the reconstruction of acoustic frames from their altered counterpart, where we use a stochastic policy to alter along various dimensions: time, frequency, and magnitude. TERA can be used for speech representations extraction or fine-tuning with downstream models. We evaluate TERA on several downstream tasks, including phoneme classification, keyword spotting, speaker recognition, and speech recognition. We present a large-scale comparison of various self-supervised models. TERA achieves strong performance in the comparison by improving upon surface features and outperforming previous models. In our experiments, we study the effect of applying different alteration techniques, pre-training on more data, and pre-training on various features. We analyze different model sizes and find that smaller models are strong representation learners than larger models, while larger models are more effective for downstream fine-tuning than smaller models. Furthermore, we show the proposed method is transferable to downstream datasets not used in pre-training.

We present Mockingjay as a new speech representation learning approach, where bidirectional Transformer encoders are pre-trained on a large amount of unlabeled speech. Previous speech representation methods learn through conditioning on past frames and predicting information about future frames. Whereas Mockingjay is designed to predict the current frame through jointly conditioning on both past and future contexts. The Mockingjay representation improves performance for a wide range of downstream tasks, including phoneme classification, speaker recognition, and sentiment classification on spoken content, while outperforming other approaches. Mockingjay is empirically powerful and can be fine-tuned with downstream models, with only 2 epochs we further improve performance dramatically. In a low resource setting with only 0.1% of labeled data, we outperform the result of Mel-features that uses all 100% labeled data.

Single-channel, speaker-independent speech separation methods have recently seen great progress. However, the accuracy, latency, and computational cost of such methods remain insufficient. The majority of the previous methods have formulated the separation problem through the time-frequency representation of the mixed signal, which has several drawbacks, including the decoupling of the phase and magnitude of the signal, the suboptimality of time-frequency representation for speech separation, and the long latency in calculating the spectrograms. To address these shortcomings, we propose a fully-convolutional time-domain audio separation network (Conv-TasNet), a deep learning framework for end-to-end time-domain speech separation. Conv-TasNet uses a linear encoder to generate a representation of the speech waveform optimized for separating individual speakers. Speaker separation is achieved by applying a set of weighting functions (masks) to the encoder output. The modified encoder representations are then inverted back to the waveforms using a linear decoder. The masks are found using a temporal convolutional network (TCN) consisting of stacked 1-D dilated convolutional blocks, which allows the network to model the long-term dependencies of the speech signal while maintaining a small model size. The proposed Conv-TasNet system significantly outperforms previous time-frequency masking methods in separating two- and three-speaker mixtures. Additionally, Conv-TasNet surpasses several ideal time-frequency magnitude masks in two-speaker speech separation as evaluated by both objective distortion measures and subjective quality assessment by human listeners. Finally, Conv-TasNet has a significantly smaller model size and a shorter minimum latency, making it a suitable solution for both offline and real-time speech separation applications.

Describing images with text is a fundamental problem in vision-language research. Current studies in this domain mostly focus on single image captioning. However, in various real applications (e.g., image editing, difference interpretation, and retrieval), generating relational captions for two images, can also be very useful. This important problem has not been explored mostly due to lack of datasets and effective models. To push forward the research in this direction, we first introduce a new language-guided image editing dataset that contains a large number of real image pairs with corresponding editing instructions. We then propose a new relational speaker model based on an encoder-decoder architecture with static relational attention and sequential multi-head attention. We also extend the model with dynamic relational attention, which calculates visual alignment while decoding. Our models are evaluated on our newly collected and two public datasets consisting of image pairs annotated with relationship sentences. Experimental results, based on both automatic and human evaluation, demonstrate that our model outperforms all baselines and existing methods on all the datasets.

The recent advancements in large language models (LLMs) have revolutionized the field of natural language processing, progressively broadening their scope to multimodal perception and generation. However, effectively integrating listening capabilities into LLMs poses significant challenges, particularly with respect to generalizing across varied contexts and executing complex auditory tasks. In this work, we introduce WavLLM, a robust and adaptive speech large language model with dual encoders, and a prompt-aware LoRA weight adapter, optimized by a two-stage curriculum learning approach. Leveraging dual encoders, we decouple different types of speech information, utilizing a Whisper encoder to process the semantic content of speech, and a WavLM encoder to capture the unique characteristics of the speaker's identity. Within the curriculum learning framework, WavLLM first builds its foundational capabilities by optimizing on mixed elementary single tasks, followed by advanced multi-task training on more complex tasks such as combinations of the elementary tasks. To enhance the flexibility and adherence to different tasks and instructions, a prompt-aware LoRA weight adapter is introduced in the second advanced multi-task training stage. We validate the proposed model on universal speech benchmarks including tasks such as ASR, ST, SV, ER, and also apply it to specialized datasets like Gaokao English listening comprehension set for SQA, and speech Chain-of-Thought (CoT) evaluation set. Experiments demonstrate that the proposed model achieves state-of-the-art performance across a range of speech tasks on the same model size, exhibiting robust generalization capabilities in executing complex tasks using CoT approach. Furthermore, our model successfully completes Gaokao tasks without specialized training. The codes, models, audio, and Gaokao evaluation set can be accessed at aka.ms/wavllm.

Generating a voice from a face image is crucial for developing virtual humans capable of interacting using their unique voices, without relying on pre-recorded human speech. In this paper, we propose Face-StyleSpeech, a zero-shot Text-To-Speech (TTS) synthesis model that generates natural speech conditioned on a face image rather than reference speech. We hypothesize that learning both speaker identity and prosody from a face image poses a significant challenge. To address the issue, our TTS model incorporates both a face encoder and a prosody encoder. The prosody encoder is specifically designed to model prosodic features that are not captured only with a face image, allowing the face encoder to focus solely on capturing the speaker identity from the face image. Experimental results demonstrate that Face-StyleSpeech effectively generates more natural speech from a face image than baselines, even for the face images the model has not trained. Samples are at our demo page https://face-stylespeech.github.io.

Recent advancements in large language models (LLMs) have revolutionized various domains, bringing significant progress and new opportunities. Despite progress in speech-related tasks, LLMs have not been sufficiently explored in multi-talker scenarios. In this work, we present a pioneering effort to investigate the capability of LLMs in transcribing speech in multi-talker environments, following versatile instructions related to multi-talker automatic speech recognition (ASR), target talker ASR, and ASR based on specific talker attributes such as sex, occurrence order, language, and keyword spoken. Our approach utilizes WavLM and Whisper encoder to extract multi-faceted speech representations that are sensitive to speaker characteristics and semantic context. These representations are then fed into an LLM fine-tuned using LoRA, enabling the capabilities for speech comprehension and transcription. Comprehensive experiments reveal the promising performance of our proposed system, MT-LLM, in cocktail party scenarios, highlighting the potential of LLM to handle speech-related tasks based on user instructions in such complex settings.

Existing studies on talking video generation have predominantly focused on single-person monologues or isolated facial animations, limiting their applicability to realistic multi-human interactions. To bridge this gap, we introduce MIT, a large-scale dataset specifically designed for multi-human talking video generation. To this end, we develop an automatic pipeline that collects and annotates multi-person conversational videos. The resulting dataset comprises 12 hours of high-resolution footage, each featuring two to four speakers, with fine-grained annotations of body poses and speech interactions. It captures natural conversational dynamics in multi-speaker scenario, offering a rich resource for studying interactive visual behaviors. To demonstrate the potential of MIT, we furthur propose CovOG, a baseline model for this novel task. It integrates a Multi-Human Pose Encoder (MPE) to handle varying numbers of speakers by aggregating individual pose embeddings, and an Interactive Audio Driver (IAD) to modulate head dynamics based on speaker-specific audio features. Together, these components showcase the feasibility and challenges of generating realistic multi-human talking videos, establishing MIT as a valuable benchmark for future research. The code is avalibale at: https://github.com/showlab/Multi-human-Talking-Video-Dataset.

We introduce StyleFusion-TTS, a prompt and/or audio referenced, style and speaker-controllable, zero-shot text-to-speech (TTS) synthesis system designed to enhance the editability and naturalness of current research literature. We propose a general front-end encoder as a compact and effective module to utilize multimodal inputs including text prompts, audio references, and speaker timbre references in a fully zero-shot manner and produce disentangled style and speaker control embeddings. Our novel approach also leverages a hierarchical conformer structure for the fusion of style and speaker control embeddings, aiming to achieve optimal feature fusion within the current advanced TTS architecture. StyleFusion-TTS is evaluated through multiple metrics, both subjectively and objectively. The system shows promising performance across our evaluations, suggesting its potential to contribute to the advancement of the field of zero-shot text-to-speech synthesis.

We present a two-speaker automatic speech recognition (ASR) system that combines DiCoW -- a diarization-conditioned variant of Whisper -- with DiariZen, a diarization pipeline built on top of Pyannote. We first evaluate both systems in out-of-domain (OOD) multilingual scenarios without any fine-tuning. In this scenario, DiariZen consistently outperforms the baseline Pyannote diarization model, demonstrating strong generalization. Despite being fine-tuned on English-only data for target-speaker ASR, DiCoW retains solid multilingual performance, indicating that encoder modifications preserve Whisper's multilingual capabilities. We then fine-tune both DiCoW and DiariZen on the MLC-SLM challenge data. The fine-tuned DiariZen continues to outperform the fine-tuned Pyannote baseline, while DiCoW sees further gains from domain adaptation. Our final system achieves a micro-average tcpWER/CER of 16.75% and ranks second in Task 2 of the MLC-SLM challenge. Lastly, we identify several labeling inconsistencies in the training data -- such as missing speech segments and incorrect silence annotations -- which can hinder diarization fine-tuning. We propose simple mitigation strategies to address these issues and improve system robustness.

Large language models (LLMs) have made significant advancements in natural language processing and are concurrently extending the language ability to other modalities, such as speech and vision. Nevertheless, most of the previous work focuses on prompting LLMs with perception abilities like auditory comprehension, and the effective approach for augmenting LLMs with speech synthesis capabilities remains ambiguous. In this paper, we conduct a comprehensive empirical exploration of boosting LLMs with the ability to generate speech, by combining pre-trained LLM LLaMA/OPT and text-to-speech synthesis model VALL-E. We compare three integration methods between LLMs and speech synthesis models, including directly fine-tuned LLMs, superposed layers of LLMs and VALL-E, and coupled LLMs and VALL-E using LLMs as a powerful text encoder. Experimental results show that, using LoRA method to fine-tune LLMs directly to boost the speech synthesis capability does not work well, and superposed LLMs and VALL-E can improve the quality of generated speech both in speaker similarity and word error rate (WER). Among these three methods, coupled methods leveraging LLMs as the text encoder can achieve the best performance, making it outperform original speech synthesis models with a consistently better speaker similarity and a significant (10.9%) WER reduction.

Large Language Models (LLMs) have made significant progress in various downstream tasks, inspiring the development of Speech Understanding Language Models (SULMs) to enable comprehensive speech-based interactions. However, most advanced SULMs are developed by the industry, leveraging large-scale datasets and computational resources that are not readily available to the academic community. Moreover, the lack of transparency in training details creates additional barriers to further innovation. In this study, we present OSUM, an Open Speech Understanding Model designed to explore the potential of training SLUMs under constrained academic resources. The OSUM model combines a Whisper encoder with a Qwen2 LLM and supports a wide range of speech tasks, including speech recognition (ASR), speech recognition with timestamps (SRWT), vocal event detection (VED), speech emotion recognition (SER), speaking style recognition (SSR), speaker gender classification (SGC), speaker age prediction (SAP), and speech-to-text chat (STTC). By employing an ASR+X training strategy, OSUM achieves efficient and stable multi-task training by simultaneously optimizing ASR alongside target tasks. Beyond delivering strong performance, OSUM emphasizes transparency by providing openly available data preparation and training methodologies, offering valuable insights and practical guidance for the academic community. By doing so, we aim to accelerate research and innovation in advanced SULM technologies.

We present a novel approach to adapting pre-trained large language models (LLMs) to perform question answering (QA) and speech continuation. By endowing the LLM with a pre-trained speech encoder, our model becomes able to take speech inputs and generate speech outputs. The entire system is trained end-to-end and operates directly on spectrograms, simplifying our architecture. Key to our approach is a training objective that jointly supervises speech recognition, text continuation, and speech synthesis using only paired speech-text pairs, enabling a `cross-modal' chain-of-thought within a single decoding pass. Our method surpasses existing spoken language models in speaker preservation and semantic coherence. Furthermore, the proposed model improves upon direct initialization in retaining the knowledge of the original LLM as demonstrated through spoken QA datasets. Audio samples can be found at https://michelleramanovich.github.io/spectron/spectron

Recent work has shown that it is possible to resynthesize high-quality speech based, not on text, but on low bitrate discrete units that have been learned in a self-supervised fashion and can therefore capture expressive aspects of speech that are hard to transcribe (prosody, voice styles, non-verbal vocalization). The adoption of these methods is still limited by the fact that most speech synthesis datasets are read, severely limiting spontaneity and expressivity. Here, we introduce Expresso, a high-quality expressive speech dataset for textless speech synthesis that includes both read speech and improvised dialogues rendered in 26 spontaneous expressive styles. We illustrate the challenges and potentials of this dataset with an expressive resynthesis benchmark where the task is to encode the input in low-bitrate units and resynthesize it in a target voice while preserving content and style. We evaluate resynthesis quality with automatic metrics for different self-supervised discrete encoders, and explore tradeoffs between quality, bitrate and invariance to speaker and style. All the dataset, evaluation metrics and baseline models are open source

This paper proposes a generative pretraining foundation model for high-quality speech restoration tasks. By directly operating on complex-valued short-time Fourier transform coefficients, our model does not rely on any vocoders for time-domain signal reconstruction. As a result, our model simplifies the synthesis process and removes the quality upper-bound introduced by any mel-spectrogram vocoder compared to prior work SpeechFlow. The proposed method is evaluated on multiple speech restoration tasks, including speech denoising, bandwidth extension, codec artifact removal, and target speaker extraction. In all scenarios, finetuning our pretrained model results in superior performance over strong baselines. Notably, in the target speaker extraction task, our model outperforms existing systems, including those leveraging SSL-pretrained encoders like WavLM. The code and the pretrained checkpoints are publicly available in the NVIDIA NeMo framework.

State-of-the-art text-to-speech (TTS) systems realize high naturalness in monolingual environments, synthesizing speech with correct multilingual accents (especially for Indic languages) and context-relevant emotions still poses difficulty owing to cultural nuance discrepancies in current frameworks. This paper introduces a new TTS architecture integrating accent along with preserving transliteration with multi-scale emotion modelling, in particularly tuned for Hindi and Indian English accent. Our approach extends the Parler-TTS model by integrating A language-specific phoneme alignment hybrid encoder-decoder architecture, and culture-sensitive emotion embedding layers trained on native speaker corpora, as well as incorporating a dynamic accent code switching with residual vector quantization. Quantitative tests demonstrate 23.7% improvement in accent accuracy (Word Error Rate reduction from 15.4% to 11.8%) and 85.3% emotion recognition accuracy from native listeners, surpassing METTS and VECL-TTS baselines. The novelty of the system is that it can mix code in real time - generating statements such as "Namaste, let's talk about <Hindi phrase>" with uninterrupted accent shifts while preserving emotional consistency. Subjective evaluation with 200 users reported a mean opinion score (MOS) of 4.2/5 for cultural correctness, much better than existing multilingual systems (p<0.01). This research makes cross-lingual synthesis more feasible by showcasing scalable accent-emotion disentanglement, with direct application in South Asian EdTech and accessibility software.

While autoregressive speech token generation models produce speech with remarkable variety and naturalness, their inherent lack of controllability often results in issues such as hallucinations and undesired vocalizations that do not conform to conditioning inputs. We introduce Koel-TTS, a suite of enhanced encoder-decoder Transformer TTS models that address these challenges by incorporating preference alignment techniques guided by automatic speech recognition and speaker verification models. Additionally, we incorporate classifier-free guidance to further improve synthesis adherence to the transcript and reference speaker audio. Our experiments demonstrate that these optimizations significantly enhance target speaker similarity, intelligibility, and naturalness of synthesized speech. Notably, Koel-TTS directly maps text and context audio to acoustic tokens, and on the aforementioned metrics, outperforms state-of-the-art TTS models, despite being trained on a significantly smaller dataset. Audio samples and demos are available on our website.

This paper describes the Microsoft end-to-end neural text to speech (TTS) system: DelightfulTTS for Blizzard Challenge 2021. The goal of this challenge is to synthesize natural and high-quality speech from text, and we approach this goal in two perspectives: The first is to directly model and generate waveform in 48 kHz sampling rate, which brings higher perception quality than previous systems with 16 kHz or 24 kHz sampling rate; The second is to model the variation information in speech through a systematic design, which improves the prosody and naturalness. Specifically, for 48 kHz modeling, we predict 16 kHz mel-spectrogram in acoustic model, and propose a vocoder called HiFiNet to directly generate 48 kHz waveform from predicted 16 kHz mel-spectrogram, which can better trade off training efficiency, modelling stability and voice quality. We model variation information systematically from both explicit (speaker ID, language ID, pitch and duration) and implicit (utterance-level and phoneme-level prosody) perspectives: 1) For speaker and language ID, we use lookup embedding in training and inference; 2) For pitch and duration, we extract the values from paired text-speech data in training and use two predictors to predict the values in inference; 3) For utterance-level and phoneme-level prosody, we use two reference encoders to extract the values in training, and use two separate predictors to predict the values in inference. Additionally, we introduce an improved Conformer block to better model the local and global dependency in acoustic model. For task SH1, DelightfulTTS achieves 4.17 mean score in MOS test and 4.35 in SMOS test, which indicates the effectiveness of our proposed system

Advances in voice conversion and text-to-speech synthesis have made automatic speaker verification (ASV) systems more susceptible to spoofing attacks. This work explores modest refinements to the AASIST anti-spoofing architecture. It incorporates a frozen Wav2Vec 2.0 encoder to retain self-supervised speech representations in limited-data settings, substitutes the original graph attention block with a standardized multi-head attention module using heterogeneous query projections, and replaces heuristic frame-segment fusion with a trainable, context-aware integration layer. When evaluated on the ASVspoof 5 corpus, the proposed system reaches a 7.6\% equal error rate (EER), improving on a re-implemented AASIST baseline under the same training conditions. Ablation experiments suggest that each architectural change contributes to the overall performance, indicating that targeted adjustments to established models may help strengthen speech deepfake detection in practical scenarios. The code is publicly available at https://github.com/KORALLLL/AASIST_SCALING.

This paper proposes RefXVC, a method for cross-lingual voice conversion (XVC) that leverages reference information to improve conversion performance. Previous XVC works generally take an average speaker embedding to condition the speaker identity, which does not account for the changing timbre of speech that occurs with different pronunciations. To address this, our method uses both global and local speaker embeddings to capture the timbre changes during speech conversion. Additionally, we observed a connection between timbre and pronunciation in different languages and utilized this by incorporating a timbre encoder and a pronunciation matching network into our model. Furthermore, we found that the variation in tones is not adequately reflected in a sentence, and therefore, we used multiple references to better capture the range of a speaker's voice. The proposed method outperformed existing systems in terms of both speech quality and speaker similarity, highlighting the effectiveness of leveraging reference information in cross-lingual voice conversion. The converted speech samples can be found on the website: http://refxvc.dn3point.com

Non-parallel many-to-many voice conversion remains an interesting but challenging speech processing task. Recently, AutoVC, a conditional autoencoder based method, achieved excellent conversion results by disentangling the speaker identity and the speech content using information-constraining bottlenecks. However, due to the pure autoencoder training method, it is difficult to evaluate the separation effect of content and speaker identity. In this paper, a novel voice conversion framework, named boldsymbol Text boldsymbol Guided boldsymbol AutoVC(TGAVC), is proposed to more effectively separate content and timbre from speech, where an expected content embedding produced based on the text transcriptions is designed to guide the extraction of voice content. In addition, the adversarial training is applied to eliminate the speaker identity information in the estimated content embedding extracted from speech. Under the guidance of the expected content embedding and the adversarial training, the content encoder is trained to extract speaker-independent content embedding from speech. Experiments on AIShell-3 dataset show that the proposed model outperforms AutoVC in terms of naturalness and similarity of converted speech.

We introduce an approach to multilingual speech synthesis which uses the meta-learning concept of contextual parameter generation and produces natural-sounding multilingual speech using more languages and less training data than previous approaches. Our model is based on Tacotron 2 with a fully convolutional input text encoder whose weights are predicted by a separate parameter generator network. To boost voice cloning, the model uses an adversarial speaker classifier with a gradient reversal layer that removes speaker-specific information from the encoder. We arranged two experiments to compare our model with baselines using various levels of cross-lingual parameter sharing, in order to evaluate: (1) stability and performance when training on low amounts of data, (2) pronunciation accuracy and voice quality of code-switching synthesis. For training, we used the CSS10 dataset and our new small dataset based on Common Voice recordings in five languages. Our model is shown to effectively share information across languages and according to a subjective evaluation test, it produces more natural and accurate code-switching speech than the baselines.

Metaverse expands the physical world to a new dimension, and the physical environment and Metaverse environment can be directly connected and entered. Voice is an indispensable communication medium in the real world and Metaverse. Fusion of the voice with environment effects is important for user immersion in Metaverse. In this paper, we proposed using the voice conversion based method for the conversion of target environment effect speech. The proposed method was named MetaSpeech, which introduces an environment effect module containing an effect extractor to extract the environment information and an effect encoder to encode the environment effect condition, in which gradient reversal layer was used for adversarial training to keep the speech content and speaker information while disentangling the environmental effects. From the experiment results on the public dataset of LJSpeech with four environment effects, the proposed model could complete the specific environment effect conversion and outperforms the baseline methods from the voice conversion task.

We present a textless speech-to-speech translation (S2ST) system that can translate speech from one language into another language and can be built without the need of any text data. Different from existing work in the literature, we tackle the challenge in modeling multi-speaker target speech and train the systems with real-world S2ST data. The key to our approach is a self-supervised unit-based speech normalization technique, which finetunes a pre-trained speech encoder with paired audios from multiple speakers and a single reference speaker to reduce the variations due to accents, while preserving the lexical content. With only 10 minutes of paired data for speech normalization, we obtain on average 3.2 BLEU gain when training the S2ST model on the VoxPopuli S2ST dataset, compared to a baseline trained on un-normalized speech target. We also incorporate automatically mined S2ST data and show an additional 2.0 BLEU gain. To our knowledge, we are the first to establish a textless S2ST technique that can be trained with real-world data and works for multiple language pairs. Audio samples are available at https://facebookresearch.github.io/speech_translation/textless_s2st_real_data/index.html .