Method and Apparatus for Spectrotemporally Improved Spectral Gap Filling in Audio Coding Using Different Noise Filling Methods

Simple SummaryContent extracted from patent full text and abstract with AI.

This invention describes improved methods and devices for filling spectral gaps (regions of missing or zeroed frequencies) in audio signals during encoding and decoding. It introduces a new way to generate and adjust the artificial content (noise or copied frequencies) used to fill these gaps, using adaptive parameters like noise, tilt, filtering, and prediction based on the content's characteristics. The approach enables more accurate and perceptually pleasing audio reconstruction, especially at low bit rates, by selecting and customizing gap-filling techniques in real time depending on detected signal features such as transientness or tonality.

Use CasesContent extracted from patent full text and abstract with AI.

• Audio codecs for streaming (music, podcasts, speech) where bandwidth efficiency and perceptual quality at low bit rates are important.
• Mobile voice and audio communications (e.g., VoIP, 5G/4G telephony, conferencing) to improve naturalness and intelligibility.
• Digital broadcasting and satellite radio codecs, improving quality within limited bitrates.
• Hearing aids or assistive listening devices, delivering better sound at low data rates.
• Archival or compression of audio files for storage and transmission.
• Any audio playback device (phones, tablets, TVs, smart speakers) that needs to reconstruct compressed audio at high quality even when some spectral information is missing.

BenefitsContent extracted from patent full text and abstract with AI.

• Reduces audible artifacts like 'holes' or unnatural noise in decoded audio, especially at low bit rates.
• Minimizes overhead and complexity in the audio encoding/decoding process by reducing extra signaling compared to previous methods.
• Automatically adapts the gap-filling strategy based on the signal's properties (e.g., whether a segment is tonal or noisy, transient or stationary), improving quality across a wide variety of audio content.
• Enables efficient codec designs for bandwidth-constrained environments, maintaining high user-perceived audio quality.
• Offers a flexible framework that can switch between different gap-filling techniques (noise, frequency prediction, copying, etc.) in real time, based on signal analysis.
• Can be implemented in both hardware and software codecs, and is compatible with modern transform-domain audio codecs (e.g., MDCT-based).

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

Embodiments according to the invention are related to methods and apparatuses for spectrotemporally improved spectral gap filling in audio coding using a filtering. Embodiments according to the invention are related to methods and apparatuses for spectrotemporally improved spectral gap filling in audio coding using different noise filling methods. Embodiments according to the invention are related to methods and apparatuses for spectrotemporally improved spectral gap filling in audio coding using a tilt.