Towards Unsupervised Phone And Word Segmentation Using Self-supervised Vector-quantized Neural Networks

Abstract

We investigate segmenting and clustering speech into low-bitrate phone-like sequences without supervision. We specifically constrain pretrained self-supervised vector-quantized (VQ) neural networks so that blocks of contiguous feature vectors are assigned to the same code, thereby giving a variable-rate segmentation of the speech into discrete units. Two segmentation methods are considered. In the first, features are greedily merged until a prespecified number of segments are reached. The second uses dynamic programming to optimize a squared error with a penalty term to encourage fewer but longer segments. We show that these VQ segmentation methods can be used without alteration across a wide range of tasks: unsupervised phone segmentation, ABX phone discrimination, same-different word discrimination, and as inputs to a symbolic word segmentation algorithm. The penalized dynamic programming method generally performs best. While performance on individual tasks is only comparable to the

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keykamper2020towards

Related papers

• Word Segmentation On Discovered Phone Units With Dynamic Programming And Self-supervised Scoring (2022)9.23
• Vq-wav2vec: Self-supervised Learning Of Discrete Speech Representations (2019)0.00
• Learning Disentangled Phone And Speaker Representations In A Semi-supervised VQ-VAE Paradigm (2020)8.09
• Unsupervised Speech Segmentation And Variable Rate Representation Learning Using Segmental Contrastive Predictive Coding (2021)9.92
• Speech Enhancement Using Self-supervised Pre-trained Model And Vector Quantization (2022)6.34
• Robust Training Of Vector Quantized Bottleneck Models (2020)11.29
• Self-supervised Learning With Random-projection Quantizer For Speech Recognition (2022)0.00
• Vector-quantized Neural Networks For Acoustic Unit Discovery In The Zerospeech 2020 Challenge (2020)13.50