Predicting Speaking Intent to Mitigate Teleconference Speech Collision

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Teleconferencing experiences significant challenges with speech collision due to latency in real-time communication, leading to awkwardness and inefficiency as participants often speak simultaneously, as visual cues used in in-person interactions are delayed or not fully facilitated.

Innovation Solution

Implementing a participant computing device with sensors (like accelerometers and gyroscopes) to predict speaking intent through machine-learned models, allowing for early indication of speaking intent to other participants, thereby reducing the likelihood of simultaneous speech.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If real-time audio transmission is used in teleconferencing, then communication efficiency is improved, but speech collision occurs due to latency

Engineering Contradiction:

Improvecommunication efficiencyVSAvoidspeech collision

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system performs preliminary detection of speaking intent using sensors (accelerometer, gyroscope, microphone) before the participant actually speaks. The machine learning model analyzes sensor data in advance to predict speaking intent, allowing the system to prepare turn-indication signals before speech occurs, thereby preventing speech collision while maintaining efficient communication flow

Inventive Principle:

Principle #10Preliminary action

2Reliability

If visual cues are used to indicate speaking intent, then speech collision is reduced, but latency in communication increases

Engineering Contradiction:

Improvespeech collision mitigationVSAvoidcommunication latency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

Each participant's computing device performs self-service by locally processing sensor data through the machine learning model to determine speaking intent. The device independently generates and transmits turn-indication signals without requiring constant server mediation, reducing communication latency while maintaining reliable speech collision mitigation through distributed intelligence

Inventive Principle:

Principle #25Self-service

3Measurement precision

If sensor-based intent detection is implemented, then speaking intent prediction accuracy is improved, but device complexity increases

Engineering Contradiction:

Improvespeaking intent detection accuracyVSAvoidsensor processing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system merges multiple sensor types (accelerometer, gyroscope, microphone) into a unified detection framework where their data is combined and processed by a single machine learning model. This integration approach improves speaking intent detection accuracy by leveraging complementary sensor information while managing device complexity through centralized processing rather than separate independent systems

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS20240339116A1Mitigating Speech Collision by Predicting Speaking Intent for Participants

Publication Date: 2024.10.10 GOOGLE LLC

AI summary

Sensor data is obtained from one or more sensors of a participant computing device. The participant computing device and one or more other participant computing devices are connected to a teleconference orchestrated by a teleconference computing system. Based at least in part on the sensor data, a participant associated with the participant computing device is determined to intend to speak to other participants of the teleconference. Information indicating that the participant intends to speak is provided to one or more of the teleconference computing system or at least one of the one or more other participant computing devices.