Apparatus and Method for Providing an Informed Multichannel Speech Presence Probability Estimation

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

This invention describes an apparatus and method for estimating the probability that speech is present in a multichannel audio environment. The key innovation is using spatial information—such as the position, direction, and proximity of sound sources—along with data from multiple sensors (e.g., microphones, proximity detectors) to more accurately determine when speech is occurring. This allows better distinction between desired speech and other sounds (like noise or background chatter), which is essential for applications like noise reduction or audio processing in complex environments.

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

• Hands-free communication systems (e.g., smart speakers, conference phones) to improve voice detection and clarity.
• Hearing aids and assistive listening devices for enhanced speech recognition and noise reduction.
• Voice-controlled appliances, robots, or vehicles that need to identify and focus on speech commands in noisy or multi-speaker environments.
• Audio surveillance systems to distinguish between speech and non-speech events more reliably.
• Video conferencing and telepresence systems for automatic speaker tracking and noise filtering.
• Recording devices for live events, meetings, or interviews where isolating speech is important.
• Smart home systems that activate on speech intent while ignoring other sounds.

BenefitsContent extracted from patent full text and abstract with AI.

• More accurate detection of speech in the presence of noise or multiple simultaneous sound sources.
• Improved noise suppression and speech enhancement by leveraging spatial context and additional sensor information.
• Reduces false positives (mistaking non-speech for speech) and false negatives (missing actual speech), leading to better user experiences.
• Highly adaptable to various acoustic environments—indoor, outdoor, multiple speakers, etc.
• Can be integrated into existing audio processing systems with minimal additional hardware.
• Enables more intelligent and selective audio filtering for clarity and intelligibility of communication or recordings.
• Supports advanced functions like speaker localization, focus, and environment-aware filtering.

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

An apparatus for providing a speech probability estimation is provided. The apparatus comprises a first speech probability estimator (110) for estimating speech probability information indicating a first probability on whether a sound field of a scene comprises speech or on whether the sound field of the scene does not comprise speech. Moreover, the apparatus comprises an output interface (120) for outputting the speech probability estimation depending on the speech probability information. The first speech probability estimator (110) is configured to estimate the first speech probability information based on at least spatial information about the sound field or spatial information on the scene.