Noro: A Noise-Robust One-shot Voice Conversion System with Hidden Speaker Representation Capabilities

One-shot voice conversion (VC) aims to alter the timbre of speech from a source speaker to match that of a target speaker using just a single reference speech from the target, while preserving the semantic content of the original source speech. Despite advancements in one-shot VC, its effectiveness decreases in real-world scenarios where reference speeches, often sourced from the internet, contain various disturbances like background noise. To address this issue, we introduce Noro, a Noise Robust One-shot VC system. Noro features innovative components tailored for VC using noisy reference speeches, including a dual-branch reference encoding module and a noise-agnostic contrastive speaker loss. Experimental results demonstrate that Noro outperforms our baseline system in both clean and noisy scenarios, highlighting its efficacy for real-world applications. Additionally, we investigate the hidden speaker representation capabilities of our baseline system by repurposing its reference encoder as a speaker encoder. The results shows that it is competitive with several advanced self-supervised learning models for speaker representation under the SUPERB settings, highlighting the potential for advancing speaker representation learning through one-shot VC task.

Speech Separation using Neural Audio Codecs with Embedding Loss

Neural audio codecs have revolutionized audio processing by enabling speech tasks to be performed on highly compressed representations. Recent work has shown that speech separation can be achieved within these compressed domains, offering faster training and reduced inference costs. However, current approaches still rely on waveform-based loss functions, necessitating unnecessary decoding steps during training. We propose a novel embedding loss for neural audio codec-based speech separation that operates directly on compressed audio representations, eliminating the need for decoding during training. To validate our approach, we conduct comprehensive evaluations using both objective metrics and perceptual assessment techniques, including intrusive and non-intrusive methods. Our results demonstrate that embedding loss can be used to train codec-based speech separation models with a 2x improvement in training speed and computational cost while achieving better DNSMOS and STOI performance on the WSJ0-2mix dataset across 3 different pre-trained codecs.

NTU-NPU System for Voice Privacy 2024 Challenge

In this work, we describe our submissions for the Voice Privacy Challenge 2024. Rather than proposing a novel speech anonymization system, we enhance the provided baselines to meet all required conditions and improve evaluated metrics. Specifically, we implement emotion embedding and experiment with WavLM and ECAPA2 speaker embedders for the B3 baseline. Additionally, we compare different speaker and prosody anonymization techniques. Furthermore, we introduce Mean Reversion F0 for B5, which helps to enhance privacy without a loss in utility. Finally, we explore disentanglement models, namely $\beta$-VAE and NaturalSpeech3 FACodec.

Bridging Speech and Text: Enhancing ASR with Pinyin-to-Character Pre-training in LLMs

The integration of large language models (LLMs) with pre-trained speech models has opened up new avenues in automatic speech recognition (ASR). While LLMs excel in multimodal understanding tasks, effectively leveraging their capabilities for ASR remains a significant challenge. This paper presents a novel training approach to enhance LLM performance in ASR tasks. We propose pre-training LLMs on Pinyin embedding sequences, which represent pronunciation features, to generate corresponding Chinese characters. This step enables the LLM to adapt to generating text from pronunciation features before encountering real speech data. Furthermore, we fine-tune the LoRA parameters to enhance the LLM's understanding of speech modality information. In AISHELL-1 corpus, our approach yields a 9.5% relative improvement in ASR tasks compared to the baseline without Pinyi-to-Character pre-training. Additionally, incorporating auxiliary text data for Pinyi-to-Character pre-training further boosts performance, achieving a 19.0% relative improvement.

Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging, and Emotion Recognition

Given recent advances in generative AI technology, a key question is how large language models (LLMs) can enhance acoustic modeling tasks using text decoding results from a frozen, pretrained automatic speech recognition (ASR) model. To explore new capabilities in language modeling for speech processing, we introduce the generative speech transcription error correction (GenSEC) challenge. This challenge comprises three post-ASR language modeling tasks: (i) post-ASR transcription correction, (ii) speaker tagging, and (iii) emotion recognition. These tasks aim to emulate future LLM-based agents handling voice-based interfaces while remaining accessible to a broad audience by utilizing open pretrained language models or agent-based APIs. We also discuss insights from baseline evaluations, as well as lessons learned for designing future evaluations.

Continual Learning Optimizations for Auto-regressive Decoder of Multilingual ASR systems

Continual Learning (CL) involves fine-tuning pre-trained models with new data while maintaining the performance on the pre-trained data. This is particularly relevant for expanding multilingual ASR (MASR) capabilities. However, existing CL methods, mainly designed for computer vision and reinforcement learning tasks, often yield sub-optimal results when directly applied to MASR. We hypothesise that this is because CL of the auto-regressive decoder in the MASR model is difficult. To verify this, we propose four optimizations on the decoder. They include decoder-layer gradient surgery, freezing unused token embeddings, suppressing output of newly added tokens, and learning rate re-scaling. Our experiments on adapting Whisper to 10 unseen languages from the Common Voice dataset demonstrate that these optimizations reduce the Average Word Error Rate (AWER) of pretrained languages from 14.2% to 12.4% compared with Experience Replay, without compromising the AWER of new languages.

Robust Zero-Shot Text-to-Speech Synthesis with Reverse Inference Optimization

In this paper, we propose reverse inference optimization (RIO), a simple and effective method designed to enhance the robustness of autoregressive-model-based zero-shot text-to-speech (TTS) systems using reinforcement learning from human feedback (RLHF). To assess the quality of speech produced by the TTS system without human annotations, RIO introduces a novel concept termed as reverse inference based on the Bayesian principle, which suggests that a high-quality generated speech should be able to be used as a prompt for subsequent generation using the same TTS model. By leveraging reverse inference as the standard to select exemplars used in RLHF from the speech samples generated by the TTS system itself, RIO steers the subsequent optimization towards a direction of enhancing the TTS robustness. The RIO framework, comprising sampling, automatic annotating, and learning, obviates the need for a reward model or pairwise preference data, and significantly improves the stability of zero-shot TTS performance by reducing the discrepancies between training and inference conditions. Our experimental results verify that RIO can effectively improve both subjective and objective metrics, including mean opinion scores, word error rates, and speaker similarity. Remarkably, RIO can also diminish the incidence of bad outputs to nearly zero percent, rivalling the robustness when using ground-truth speech as the prompt.

Temporal-Channel Modeling in Multi-head Self-Attention for Synthetic Speech Detection

Recent synthetic speech detectors leveraging the Transformer model have superior performance compared to the convolutional neural network counterparts. This improvement could be due to the powerful modeling ability of the multi-head self-attention (MHSA) in the Transformer model, which learns the temporal relationship of each input token. However, artifacts of synthetic speech can be located in specific regions of both frequency channels and temporal segments, while MHSA neglects this temporal-channel dependency of the input sequence. In this work, we proposed a Temporal-Channel Modeling (TCM) module to enhance MHSA's capability for capturing temporal-channel dependencies. Experimental results on the ASVspoof 2021 show that with only 0.03M additional parameters, the TCM module can outperform the state-of-the-art system by 9.25% in EER. Further ablation study reveals that utilizing both temporal and channel information yields the most improvement for detecting synthetic speech.

Towards Audio Codec-based Speech Separation

Recent improvements in neural audio codec (NAC) models have generated interest in adopting pre-trained codecs for a variety of speech processing applications to take advantage of the efficiencies gained from high compression, but these have yet been applied to the speech separation (SS) task. SS can benefit from high compression because the compute required for traditional SS models makes them impractical for many edge computing use cases. However, SS is a waveform-masking task where compression tends to introduce distortions that severely impact performance. Here we propose a novel task of Audio Codec-based SS, where SS is performed within the embedding space of a NAC, and propose a new model, Codecformer, to address this task. At inference, Codecformer achieves a 52x reduction in MAC while producing separation performance comparable to a cloud deployment of Sepformer. This method charts a new direction for performing efficient SS in practical scenarios.

Dataset-Distillation Generative Model for Speech Emotion Recognition

Deep learning models for speech rely on large datasets, presenting computational challenges. Yet, performance hinges on training data size. Dataset Distillation (DD) aims to learn a smaller dataset without much performance degradation when training with it. DD has been investigated in computer vision but not yet in speech. This paper presents the first approach for DD to speech targeting Speech Emotion Recognition on IEMOCAP. We employ Generative Adversarial Networks (GANs) not to mimic real data but to distil key discriminative information of IEMOCAP that is useful for downstream training. The GAN then replaces the original dataset and can sample custom synthetic dataset sizes. It performs comparably when following the original class imbalance but improves performance by 0.3% absolute UAR with balanced classes. It also reduces dataset storage and accelerates downstream training by 95% in both cases and reduces speaker information which could help for a privacy application.