Unsupervised speech representation learning using WaveNet autoencoders

Abstract

We consider the task of unsupervised extraction of meaningful latent representations of speech by applying autoencoding neural networks to speech waveforms. The goal is to learn a representation able to capture high level semantic content from the signal, e.g.\ phoneme identities, while being invariant to confounding low level details in the signal such as the underlying pitch contour or background noise. Since the learned representation is tuned to contain only phonetic content, we resort to using a high capacity WaveNet decoder to infer information discarded by the encoder from previous samples. Moreover, the behavior of autoencoder models depends on the kind of constraint that is applied to the latent representation. We compare three variants: a simple dimensionality reduction bottleneck, a Gaussian Variational Autoencoder (VAE), and a discrete Vector Quantized VAE (VQ-VAE). We analyze the quality of learned representations in terms of speaker independence, the ability to predict phonetic content, and the ability to accurately reconstruct individual spectrogram frames. Moreover, for discrete encodings extracted using the VQ-VAE, we measure the ease of mapping them to phonemes. We introduce a regularization scheme that forces the representations to focus on the phonetic content of the utterance and report performance comparable with the top entries in the ZeroSpeech 2017 unsupervised acoustic unit discovery task.

arXiv commentAccepted to IEEE TASLP, final version available at http://dx.doi.org/10.1109/TASLP.2019.2938863

References in corpus (3)

• Neural Discrete Representation Learning
• A segmental framework for fully-unsupervised large-vocabulary speech recognition
• Unsupervised Incremental Learning and Prediction of Music Signals

Cited by in corpus (40)

• WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing
• A Unifying Review of Deep and Shallow Anomaly Detection
• Deep Learning for Deepfakes Creation and Detection: A Survey
• Self-Supervised Speech Representation Learning: A Review
• Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders
• TERA: Self-Supervised Learning of Transformer Encoder Representation for Speech
• BigSSL: Exploring the Frontier of Large-Scale Semi-Supervised Learning for Automatic Speech Recognition
• Deep Contextualized Acoustic Representations For Semi-Supervised Speech Recognition
• Data Augmenting Contrastive Learning of Speech Representations in the Time Domain
• CiwGAN and fiwGAN: Encoding information in acoustic data to model lexical learning with Generative Adversarial Networks
• Robust Training of Vector Quantized Bottleneck Models
• Unsupervised Representation Disentanglement using Cross Domain Features and Adversarial Learning in Variational Autoencoder based Voice Conversion
• Self-supervised language learning from raw audio: Lessons from the Zero Resource Speech Challenge
• WaveNet-Based Deep Neural Networks for the Characterization of Anomalous Diffusion (WADNet)
• Does Visual Self-Supervision Improve Learning of Speech Representations for Emotion Recognition?
• Scalable and Efficient Neural Speech Coding: A Hybrid Design
• Word Segmentation on Discovered Phone Units with Dynamic Programming and Self-Supervised Scoring
• TGAVC: Improving Autoencoder Voice Conversion with Text-Guided and Adversarial Training
• Supervised and Self-supervised Pretraining Based COVID-19 Detection Using Acoustic Breathing/Cough/Speech Signals
• Efficient Rare Event Sampling with Unsupervised Normalising Flows
• Learning minimal representations of stochastic processes with variational autoencoders
• Autovocoder: Fast Waveform Generation from a Learned Speech Representation using Differentiable Digital Signal Processing
• Interpreting intermediate convolutional layers of generative CNNs trained on waveforms
• Exploiting Cross-Lingual Speaker and Phonetic Diversity for Unsupervised Subword Modeling
• Mask-GVAE: Blind Denoising Graphs via Partition
• End2End Acoustic to Semantic Transduction
• Parallel Synthesis for Autoregressive Speech Generation
• Generative Models for Improved Naturalness, Intelligibility, and Voicing of Whispered Speech
• OCTOPUS: Overcoming Performance andPrivatization Bottlenecks in Distributed Learning
• StarGAN-ZSVC: Towards Zero-Shot Voice Conversion in Low-Resource Contexts
• Interpreting intermediate convolutional layers in unsupervised acoustic word classification
• A Hierarchical Subspace Model for Language-Attuned Acoustic Unit Discovery
• Word-level Embeddings for Cross-Task Transfer Learning in Speech Processing
• Privacy-oriented manipulation of speaker representations
• Identity-Based Patterns in Deep Convolutional Networks: Generative Adversarial Phonology and Reduplication
• Comparison of Self-Supervised Speech Pre-Training Methods on Flemish Dutch
• Unsupervised Multimodal Word Discovery based on Double Articulation Analysis with Co-occurrence cues
• Addressing Data Scarcity in Multimodal User State Recognition by Combining Semi-Supervised and Supervised Learning
• Guided contrastive self-supervised pre-training for automatic speech recognition
• The effectiveness of unsupervised subword modeling with autoregressive and cross-lingual phone-aware networks