Phoneme Sequence Analysis for Audio Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for measuring similarity between spoken audio segments are inefficient due to inaccurate and slow speech-to-text algorithms, and time-consuming human transcription, while low-level feature-extraction approaches are inadequate for identifying similarities between audio files.
Innovation Solution
A method using a computer analysis module to determine the frequency and weighted frequency of phoneme sequences in audio sources, comparing these to generate a similarity score, and employing techniques like cosine angle calculation and latent semantic analysis to assess phonetic similarity between audio segments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If speech-to-text algorithms are used to measure similarity between audio segments, then text transcripts can be generated, but the process is inaccurate and slow
Solution Approach 1:
The patent extracts phoneme sequences from audio signals directly using phoneme-based feature extraction, bypassing the need for speech-to-text conversion. This extraction approach focuses on retrieving phonetic content without requiring accurate word-level transcription, thereby achieving similarity measurement without the speed and accuracy limitations of speech-to-text algorithms
Solution Approach 2:
The patent introduces phoneme sequences as an intermediary representation between raw audio and text transcripts. By comparing phoneme sequences instead of full transcripts, the system achieves accurate similarity measurement while avoiding the computational burden and inaccuracy of complete speech-to-text conversion
2Measurement precision
If human transcription is used to measure similarity between audio segments, then accurate text transcripts can be obtained, but the process is time-consuming and expensive
Solution Approach 1:
The patent enables the system to perform phoneme sequence extraction and comparison automatically without human intervention. The phoneme-based approach allows computers to directly analyze and compare audio segments through automated feature extraction and similarity calculation, eliminating time-consuming and expensive human transcription while maintaining measurement accuracy
Solution Approach 2:
The patent replaces the mechanical process of human transcription with automated phoneme-based feature extraction and comparison. By substituting human linguistic analysis with computational phoneme sequence analysis, the system achieves accurate similarity measurement instantaneously without the time and cost constraints of manual transcription
3Productivity
If low-level feature-extraction approaches are used to identify similarities between audio files, then processing speed can be improved, but the approach is inadequate for identifying phonetic similarities
Solution Approach 1:
The patent changes the parameter of feature extraction from low-level acoustic features to phoneme-level features. By extracting and comparing phoneme sequences instead of basic acoustic characteristics, the system maintains processing efficiency while significantly improving the ability to detect phonetic similarities between audio segments
Data Source
AI summary
A method for determining a similarity between a first audio source and a second audio source includes: for the first audio source, determining a first frequency of occurrence for each of a plurality of phoneme sequences and determining a first weighted frequency for each of the plurality of phoneme sequences based on the first frequency of occurrence for the phoneme sequence; for the second audio source, determining a second frequency of occurrence for each of a plurality of phoneme sequences and determining a second weighted frequency for each of the plurality of phoneme sequences based on the second frequency of occurrence for the phoneme sequence; comparing the first weighted frequency for each phoneme sequence with the second weighted frequency for the corresponding phoneme sequence; and generating a similarity score representative of a similarity between the first audio source and the second audio source based on the results of the comparing.