Real-Time Sound Quality Prediction Using Convolutional Neural Networks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Amateur voice recordings often suffer from poor sound quality due to suboptimal room acoustics and excessive background noise, with existing solutions providing inadequate real-time feedback to improve recording setups.

Innovation Solution

A system that uses a convolutional neural network to predict speech transmission index and signal-to-noise ratio in real-time, providing visual feedback on sound quality through a graphical user interface to help users optimize their recording environment.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If real-time sound quality analysis is implemented, then recording quality improves, but computational complexity increases

Engineering Contradiction:

Improverecording qualityVSAvoidcomputational complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary analysis by pre-processing audio data into frequency spectra and preparing quality metrics before final rendering. This allows complex acoustic analysis to be prepared in advance, reducing real-time computational burden during actual recording operations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces intermediary processing layers including frequency spectrum analysis and intermediate quality metrics that bridge raw audio input and final quality assessment. These intermediaries break down complex analysis into manageable stages, reducing overall computational complexity.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If comprehensive sound quality metrics are calculated, then feedback accuracy improves, but processing time increases

Engineering Contradiction:

Improvefeedback accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system calculates a selective subset of sound quality metrics based on recording conditions and requirements. Rather than computing all possible acoustic parameters, it focuses on the most relevant metrics for the current recording scenario, reducing processing time while maintaining sufficient feedback accuracy.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent applies different levels of analysis depth to different frequency ranges and time periods in the audio signal. Critical frequency bands receive more detailed analysis while less important regions use simplified metrics, optimizing the balance between accuracy and processing speed.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS11138989B2Sound quality prediction and interface to facilitate high-quality voice recordings

Publication Date: 2021.10.05 ADOBE INC

AI summary

Embodiments of the present invention provide systems, methods, and computer storage media for sound quality prediction and real-time feedback about sound quality, such as room acoustics quality and background noise. Audio data can be sampled from a live sound source and stored in an audio buffer. The audio data in the buffer is analyzed to calculate a stream of values of one or more sound quality measures, such as speech transmission index and signal-to-noise ratio. Speech transmission index can be calculated using a convolution neural network configured to predict speech transmission index from reverberant speech. The stream of values can be used to provide real-time feedback about sound quality of the audio data. For example, a visual indicator on a graphical user interface can be updated based on consistency of the values over time. The real-time feedback about sound quality can help users optimize their recording setup.