Neural Audio
Neural audio focuses on developing efficient and high-quality methods for representing and manipulating audio signals using neural networks, primarily aiming for improved compression and generation. Current research emphasizes advancements in neural audio codecs, often employing architectures like transformers, variational autoencoders, and diffusion models, along with novel quantization techniques such as residual vector quantization to achieve high fidelity at low bitrates. This field is significant for its potential to revolutionize audio processing applications, including speech synthesis, speech recognition, and real-time communication, by enabling more efficient storage, transmission, and manipulation of audio data.
Papers
November 29, 2024
November 1, 2024
October 30, 2024
October 29, 2024
October 18, 2024
October 14, 2024
October 6, 2024
September 28, 2024
September 26, 2024
September 21, 2024
September 20, 2024
September 19, 2024
September 18, 2024
September 17, 2024
September 9, 2024
September 6, 2024
August 19, 2024
August 13, 2024
August 12, 2024