Speech Frame

A speech frame represents a short segment of speech, typically 10-30 milliseconds, used as a fundamental unit in speech analysis and recognition. Current research focuses on improving the efficiency and accuracy of speech frame processing, particularly within self-supervised learning models like HuBERT, employing techniques such as attention map reusing and masking distillation to compress large transformer-based architectures. These advancements aim to enhance the performance of various speech technologies, including automatic speech recognition (ASR) and text-to-speech (TTS), while reducing computational demands and making them more accessible for wider applications. The development of more efficient and interpretable models for speech frame analysis is crucial for advancing the field and improving the robustness of speech-related applications.

Papers

December 14, 2023

STaR: Distilling Speech Temporal Relation for Lightweight Speech Self-Supervised Learning Models
Kangwook Jang, Sungnyun Kim, Hoirin Kim
Self Supervised Learning Temporal Relation Speech Supervised Learning Model Speech Segment Speech Transformer Speech Frame

May 19, 2023

Recycle-and-Distill: Universal Compression Strategy for Transformer-based Speech SSL Models with Attention Map Reusing and Masking Distillation
Kangwook Jang, Sungnyun Kim, Se-Young Yun, Hoirin Kim
Speech Supervised Learning Model Speech Processing Task Speech Transformer Functional Compression Masked Distillation Speech Frame

March 6, 2023

FoundationTTS: Text-to-Speech for ASR Customization with Generative Language Model
Ruiqing Xue, Yanqing Liu, Lei He, Xu Tan, Linquan Liu, Edward Lin, Sheng Zhao
Speech Recognition Text to Speech Generative Language Model Flawed Foundation Frame Level Continuous Speech Recognition Speech Frame

February 24, 2022

Ask2Mask: Guided Data Selection for Masked Speech Modeling
Murali Karthick Baskar, Andrew Rosenberg, Bhuvana Ramabhadran, Yu Zhang, Pedro Moreno
Automatic Speech Recognition ASR Model Machine Self Confidence Speech Frame

December 15, 2021

Speech frame implementation for speech analysis and recognition
A. A. Konev, V. S. Khlebnikov, A. Yu. Yakimuk
Speech Analysis Recognition Rate Speech Signal Rich Attribute Russian Language Speech Segment Speech Frame

December 12, 2021

Visualising and Explaining Deep Learning Models for Speech Quality Prediction
H. Tilkorn, G. Mittag, S. Möller
Deep Learning Model Recurrent Neural Network RNN Model Non Intrusive Speech Speech Quality Prediction Speech Frame