Peer-Aware Voice Stream Ranking for VoIP Bandwidth Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
VoIP multi-party voice conferencing faces challenges due to variable delays and packet losses in wide-area networks, and traditional silence suppression techniques are inefficient as they rely solely on individual client speech/silence tests, leading to increased load on central bridges.
Innovation Solution
The peer-aware voice stream ranking method assesses voice activity not only from a client's own audio but also from other participants, using Voice Activity Scores (VAS) with voiceness and energy level components, to determine when to send voice packets, thereby reducing bandwidth utilization and CPU processing on the bridge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional silence suppression techniques are used where each client makes decisions based solely on its own speech/silence test, then the decision-making process is simple and independent, but the bandwidth utilization is inefficient and the load on central bridges increases
Solution Approach 1:
The patent implements feedback by having clients receive voice activity information from other participants and use this feedback to adjust their own packet transmission decisions. Each client's decision to send packets is influenced by the voice activity levels of other participants, creating a coordinated system that optimizes bandwidth utilization across the entire conference.
Solution Approach 2:
The system dynamically adjusts packet transmission decisions based on real-time voice activity detection. Clients continuously monitor their own voice activity and the voice activity of other participants, and dynamically modify their transmission behavior accordingly. This dynamic adaptation allows the system to optimize bandwidth usage in response to changing speaking conditions.
2Productivity
If silence suppression is implemented to reduce bridge load, then the number of packets sent to the bridge is reduced, but the bridge must still reserve significant resources to handle worst-case packet flooding
Solution Approach 1:
The system performs preliminary voice activity detection and ranking before packets reach the bridge. By pre-determining which participants should be transmitting based on voice activity levels and peer awareness, the system reduces the number of packets that actually reach the bridge, allowing the bridge to operate more efficiently without needing to reserve excessive resources for worst-case scenarios.
3Reliability
If packet compression is performed separately for each individual client to handle variable delays and packet losses, then each client's voice is properly processed, but the load on the central bridge increases linearly with the number of clients
Solution Approach 1:
Instead of performing full packet compression and processing for all clients at the bridge, the system applies partial action by having clients perform local voice activity detection and decision-making. Only clients with significant voice activity and high ranking transmit packets, reducing the overall processing load on the bridge while maintaining reliable voice transmission for active speakers.
Data Source
AI summary
A peer-aware voice stream ranking method that makes decisions based on information about participants of a voice conference over a network. Whether to send a participant's own audio packet out on the network is based both on information about the participant's own voice packet and voice packets that the participant receives from other clients. A Voice Activity Score (VAS) is computed for each frame of a particular voice stream. The VAS includes a voiceness component, indicating the likelihood that the audio frame contains speech or voice, and an energy level component that indicating the ratio of current frame energy to the long-term average of energy for a current speaker. Using the VAS from the participants, the method also ranks the client's voice stream as compared to other clients' voice streams in the voice conference. If there are participants higher ranking, the client's voice stream is not sent.