Singular Voice Audio Stream Generation via ICA

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Devices struggle to accurately detect and interpret voice commands from a single speaker in environments with multiple speakers, leading to incorrect or missed detections due to overlapping conversations, where the length of the voice command is unknown and the device may receive unrelated audio inputs.

Innovation Solution

The system generates a singular voice audio stream using Independent Component Analysis for blind source separation of speech signals, allowing the device to isolate and recognize voice commands, including wake words, by processing audio streams from multiple microphones and providing the relevant stream to a server for interpretation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the device receives audio input from multiple microphones in a multi-speaker environment, then it can capture all spoken content, but it becomes difficult to accurately identify and interpret voice commands from a single speaker

Engineering Contradiction:
Improvevoice command detection accuracyVSAvoidinterference from unrelated audio input
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent segments the mixed audio stream into separate source streams using Independent Component Analysis (ICA). Each microphone's audio input is decomposed into statistically independent source signals, allowing the system to separate the voice command from other speakers' conversations. This segmentation enables the device to focus on the relevant speaker while filtering out interference from others.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary processing layer (ICA algorithm) between the microphones and the voice command interpretation system. This intermediary performs blind source separation to isolate the target speaker's voice from the mixed audio, acting as a mediator that transforms the problematic multi-speaker input into a clean single-speaker stream for accurate command recognition.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If the device processes all audio content from multiple speakers, then it captures complete conversation context, but it increases processing load and reduces speed of voice command recognition

Engineering Contradiction:
Improvevoice command recognition speedVSAvoidprocessing power consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent extracts only the relevant voice command stream from the mixed audio using ICA-based source separation. Instead of processing all audio content from multiple speakers, the system identifies and extracts the specific stream containing the wake word and voice command, discarding or minimizing processing of unrelated audio. This extraction significantly reduces computational load while maintaining recognition accuracy.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial processing by focusing computational resources only on the extracted voice command stream rather than analyzing the entire multi-speaker audio environment. The ICA algorithm performs the separation once, and then subsequent voice recognition processes operate only on the isolated stream, reducing overall processing requirements compared to analyzing all audio content in detail.

Inventive Principle:
Principle #16Partial or excessive action

3Reliability

If the device waits for complete speech to detect voice commands, then it ensures accurate command capture, but it increases latency and reduces user experience

Engineering Contradiction:
Improvevoice command detection reliabilityVSAvoidcommand detection latency
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent performs preliminary source separation using ICA to isolate the target speaker's stream before voice command detection begins. By pre-processing the audio to create clean, separated streams, the system enables earlier and more reliable detection of wake words and commands without waiting for complete speech segments. This preliminary action reduces the time needed for subsequent recognition while maintaining accuracy.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3866158B1Systems and methods for generating a singular voice audio stream
Publication Date: 2024.02.14 SPOTIFY
  • EP3866158B1 patent drawingFigure 1
  • EP3866158B1 patent drawingFigure 2
  • EP3866158B1 patent drawingFigure 3

AI summary

An electronic device (102) associated with a media-providing service (100) receives a first set of audio streams corresponding to a plurality of microphones. The electronic device generates a second set of audio streams from the first set of audio streams. The second set of audio streams corresponds to a plurality of independent voices and in some cases, ambient noise. The electronic device detects a beginning of a voice command to play media content from the media-providing service in a first audio stream. The electronic device also detects an end of the voice command in the first audio stream. The end of the voice command overlaps with speech in a second audio stream in the second set of audio streams. In response to detecting the voice command, the electronic device plays the media content from the media-providing service.