Language Model Can Listen While Speaking

Dialogue serves as the most natural manner of human-computer interaction (HCI). Recent advancements in speech language models (SLM) have significantly enhanced speech-based conversational AI. However, these models are limited to turn-based conversation, lacking the ability to interact with humans in real-time spoken scenarios, for example, being interrupted when the generated content is not satisfactory. To address these limitations, we explore full duplex modeling (FDM) in interactive speech language models (iSLM), focusing on enhancing real-time interaction and, more explicitly, exploring the quintessential ability of interruption. We introduce a novel model design, namely listening-while-speaking language model (LSLM), an end-to-end system equipped with both listening and speaking channels. Our LSLM employs a token-based decoder-only TTS for speech generation and a streaming self-supervised learning (SSL) encoder for real-time audio input. LSLM fuses both channels for autoregressive generation and detects turn-taking in real time. Three fusion strategies -- early fusion, middle fusion, and late fusion -- are explored, with middle fusion achieving an optimal balance between speech generation and real-time interaction. Two experimental settings, command-based FDM and voice-based FDM, demonstrate LSLM's robustness to noise and sensitivity to diverse instructions. Our results highlight LSLM's capability to achieve duplex communication with minimal impact on existing systems. This study aims to advance the development of interactive speech dialogue systems, enhancing their applicability in real-world contexts.

Seed-TTS: A Family of High-Quality Versatile Speech Generation Models

We introduce Seed-TTS, a family of large-scale autoregressive text-to-speech (TTS) models capable of generating speech that is virtually indistinguishable from human speech. Seed-TTS serves as a foundation model for speech generation and excels in speech in-context learning, achieving performance in speaker similarity and naturalness that matches ground truth human speech in both objective and subjective evaluations. With fine-tuning, we achieve even higher subjective scores across these metrics. Seed-TTS offers superior controllability over various speech attributes such as emotion and is capable of generating highly expressive and diverse speech for speakers in the wild. Furthermore, we propose a self-distillation method for speech factorization, as well as a reinforcement learning approach to enhance model robustness, speaker similarity, and controllability. We additionally present a non-autoregressive (NAR) variant of the Seed-TTS model, named $\text{Seed-TTS}_\text{DiT}$, which utilizes a fully diffusion-based architecture. Unlike previous NAR-based TTS systems, $\text{Seed-TTS}_\text{DiT}$ does not depend on pre-estimated phoneme durations and performs speech generation through end-to-end processing. We demonstrate that this variant achieves comparable performance to the language model-based variant and showcase its effectiveness in speech editing. We encourage readers to listen to demos at \url{https://bytedancespeech.github.io/seedtts_tech_report}.

U-Style: Cascading U-nets with Multi-level Speaker and Style Modeling for Zero-Shot Voice Cloning

Zero-shot speaker cloning aims to synthesize speech for any target speaker unseen during TTS system building, given only a single speech reference of the speaker at hand. Although more practical in real applications, the current zero-shot methods still produce speech with undesirable naturalness and speaker similarity. Moreover, endowing the target speaker with arbitrary speaking styles in the zero-shot setup has not been considered. This is because the unique challenge of zero-shot speaker and style cloning is to learn the disentangled speaker and style representations from only short references representing an arbitrary speaker and an arbitrary style. To address this challenge, we propose U-Style, which employs Grad-TTS as the backbone, particularly cascading a speaker-specific encoder and a style-specific encoder between the text encoder and the diffusion decoder. Thus, leveraging signal perturbation, U-Style is explicitly decomposed into speaker- and style-specific modeling parts, achieving better speaker and style disentanglement. To improve unseen speaker and style modeling ability, these two encoders conduct multi-level speaker and style modeling by skip-connected U-nets, incorporating the representation extraction and information reconstruction process. Besides, to improve the naturalness of synthetic speech, we adopt mean-based instance normalization and style adaptive layer normalization in these encoders to perform representation extraction and condition adaptation, respectively. Experiments show that U-Style significantly surpasses the state-of-the-art methods in unseen speaker cloning regarding naturalness and speaker similarity. Notably, U-Style can transfer the style from an unseen source speaker to another unseen target speaker, achieving flexible combinations of desired speaker timbre and style in zero-shot voice cloning.

DiCLET-TTS: Diffusion Model based Cross-lingual Emotion Transfer for Text-to-Speech -- A Study between English and Mandarin

While the performance of cross-lingual TTS based on monolingual corpora has been significantly improved recently, generating cross-lingual speech still suffers from the foreign accent problem, leading to limited naturalness. Besides, current cross-lingual methods ignore modeling emotion, which is indispensable paralinguistic information in speech delivery. In this paper, we propose DiCLET-TTS, a Diffusion model based Cross-Lingual Emotion Transfer method that can transfer emotion from a source speaker to the intra- and cross-lingual target speakers. Specifically, to relieve the foreign accent problem while improving the emotion expressiveness, the terminal distribution of the forward diffusion process is parameterized into a speaker-irrelevant but emotion-related linguistic prior by a prior text encoder with the emotion embedding as a condition. To address the weaker emotional expressiveness problem caused by speaker disentanglement in emotion embedding, a novel orthogonal projection based emotion disentangling module (OP-EDM) is proposed to learn the speaker-irrelevant but emotion-discriminative embedding. Moreover, a condition-enhanced DPM decoder is introduced to strengthen the modeling ability of the speaker and the emotion in the reverse diffusion process to further improve emotion expressiveness in speech delivery. Cross-lingual emotion transfer experiments show the superiority of DiCLET-TTS over various competitive models and the good design of OP-EDM in learning speaker-irrelevant but emotion-discriminative embedding.

Robust MelGAN: A robust universal neural vocoder for high-fidelity TTS

In current two-stage neural text-to-speech (TTS) paradigm, it is ideal to have a universal neural vocoder, once trained, which is robust to imperfect mel-spectrogram predicted from the acoustic model. To this end, we propose Robust MelGAN vocoder by solving the original multi-band MelGAN's metallic sound problem and increasing its generalization ability. Specifically, we introduce a fine-grained network dropout strategy to the generator. With a specifically designed over-smooth handler which separates speech signal intro periodic and aperiodic components, we only perform network dropout to the aperodic components, which alleviates metallic sounding and maintains good speaker similarity. To further improve generalization ability, we introduce several data augmentation methods to augment fake data in the discriminator, including harmonic shift, harmonic noise and phase noise. Experiments show that Robust MelGAN can be used as a universal vocoder, significantly improving sound quality in TTS systems built on various types of data.

DSPGAN: a GAN-based universal vocoder for high-fidelity TTS by time-frequency domain supervision from DSP

Recent development of neural vocoders based on the generative adversarial neural network (GAN) has shown their advantages of generating raw waveform conditioned on mel-spectrogram with fast inference speed and lightweight networks. Whereas, it is still challenging to train a universal neural vocoder that can synthesize high-fidelity speech from various scenarios with unseen speakers, languages, and speaking styles. In this paper, we propose DSPGAN, a GAN-based universal vocoder for high-fidelity speech synthesis by applying the time-frequency domain supervision from digital signal processing (DSP). To eliminate the mismatch problem caused by the ground-truth spectrograms in training phase and the predicted spectrograms in inference phase, we leverage the mel-spectrogram extracted from the waveform generated by a DSP module, rather than the predicted mel-spectrogram from the Text-to-Speech (TTS) acoustic model, as the time-frequency domain supervision to the GAN-based vocoder. We also utilize sine excitation as the time-domain supervision to improve the harmonic modeling and eliminate various artifacts of the GAN-based vocoder. Experimental results show that DSPGAN significantly outperforms the compared approaches and can generate high-fidelity speech based on diverse data in TTS.

Glow-WaveGAN 2: High-quality Zero-shot Text-to-speech Synthesis and Any-to-any Voice Conversion

The zero-shot scenario for speech generation aims at synthesizing a novel unseen voice with only one utterance of the target speaker. Although the challenges of adapting new voices in zero-shot scenario exist in both stages -- acoustic modeling and vocoder, previous works usually consider the problem from only one stage. In this paper, we extend our previous Glow-WaveGAN to Glow-WaveGAN 2, aiming to solve the problem from both stages for high-quality zero-shot text-to-speech and any-to-any voice conversion. We first build a universal WaveGAN model for extracting latent distribution $p(z)$ of speech and reconstructing waveform from it. Then a flow-based acoustic model only needs to learn the same $p(z)$ from texts, which naturally avoids the mismatch between the acoustic model and the vocoder, resulting in high-quality generated speech without model fine-tuning. Based on a continuous speaker space and the reversible property of flows, the conditional distribution can be obtained for any speaker, and thus we can further conduct high-quality zero-shot speech generation for new speakers. We particularly investigate two methods to construct the speaker space, namely pre-trained speaker encoder and jointly-trained speaker encoder. The superiority of Glow-WaveGAN 2 has been proved through TTS and VC experiments conducted on LibriTTS corpus and VTCK corpus.

AdaVITS: Tiny VITS for Low Computing Resource Speaker Adaptation

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.

NaturalSpeech: End-to-End Text to Speech Synthesis with Human-Level Quality

Text to speech (TTS) has made rapid progress in both academia and industry in recent years. Some questions naturally arise that whether a TTS system can achieve human-level quality, how to define/judge that quality and how to achieve it. In this paper, we answer these questions by first defining the human-level quality based on the statistical significance of subjective measure and introducing appropriate guidelines to judge it, and then developing a TTS system called NaturalSpeech that achieves human-level quality on a benchmark dataset. Specifically, we leverage a variational autoencoder (VAE) for end-to-end text to waveform generation, with several key modules to enhance the capacity of the prior from text and reduce the complexity of the posterior from speech, including phoneme pre-training, differentiable duration modeling, bidirectional prior/posterior modeling, and a memory mechanism in VAE. Experiment evaluations on popular LJSpeech dataset show that our proposed NaturalSpeech achieves -0.01 CMOS (comparative mean opinion score) to human recordings at the sentence level, with Wilcoxon signed rank test at p-level p >> 0.05, which demonstrates no statistically significant difference from human recordings for the first time on this dataset.

VISinger: Variational Inference with Adversarial Learning for End-to-End Singing Voice Synthesis

In this paper, we propose VISinger, a complete end-to-end high-quality singing voice synthesis (SVS) system that directly generates audio waveform from lyrics and musical score. Our approach is inspired by VITS, which adopts VAE-based posterior encoder augmented with normalizing flow-based prior encoder and adversarial decoder to realize complete end-to-end speech generation. VISinger follows the main architecture of VITS, but makes substantial improvements to the prior encoder based on the characteristics of singing. First, instead of using phoneme-level mean and variance of acoustic features, we introduce a length regulator and a frame prior network to get the frame-level mean and variance on acoustic features, modeling the rich acoustic variation in singing. Second, we further introduce an F0 predictor to guide the frame prior network, leading to stabler singing performance. Finally, to improve the singing rhythm, we modify the duration predictor to specifically predict the phoneme to note duration ratio, helped with singing note normalization. Experiments on a professional Mandarin singing corpus show that VISinger significantly outperforms FastSpeech+Neural-Vocoder two-stage approach and the oracle VITS; ablation study demonstrates the effectiveness of different contributions.