Disentangling Speech And Non-speech Components For Building Robust Acoustic Models From Found Data

Abstract

In order to build language technologies for majority of the languages, it is important to leverage the resources available in public domain on the internet - commonly referred to as `Found Data'. However, such data is characterized by the presence of non-standard, non-trivial variations. For instance, speech resources found on the internet have non-speech content, such as music. Therefore, speech recognition and speech synthesis models need to be robust to such variations. In this work, we present an analysis to show that it is important to disentangle the latent causal factors of variation in the original data to accomplish these tasks. Based on this, we present approaches to disentangle such variations from the data using Latent Stochastic Models. Specifically, we present a method to split the latent prior space into continuous representations of dominant speech modes present in the magnitude spectra of audio signals. We propose a completely unsupervised approach using multinode late

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keygurunath2019disentangling

Related papers

• Disentangling Textual And Acoustic Features Of Neural Speech Representations (2024)0.00
• Learning Disentangled Speech Representations (2023)0.00
• Disentangling Prosody Representations With Unsupervised Speech Reconstruction (2022)0.00
• Self-supervised Neural Factor Analysis For Disentangling Utterance-level Speech Representations (2023)0.00
• Unsupervised Speech Segmentation: A General Approach Using Speech Language Models (2025)2.60
• Deep Variational Generative Models For Audio-visual Speech Separation (2020)0.00
• Adversarial Feature Learning And Unsupervised Clustering Based Speech Synthesis For Found Data With Acoustic And Textual Noise (2020)7.16
• 3d-speaker: A Large-scale Multi-device, Multi-distance, And Multi-dialect Corpus For Speech Representation Disentanglement (2023)0.00