Pitch Selection Modules for Melody Transcription
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio signal processing technologies face inefficiencies in pitch selection, voicing detection, and vibrato suppression, leading to inaccuracies in transcribing sung or hummed melodies and speech processing.
Innovation Solution
A series of audio processing modules are developed to identify and remove buzz, select pitch candidates based on continuity and octave consistency, adjust pitches to align with harmonic peaks, determine voiced signals, extract harmonics, and detect vibrato, utilizing techniques such as histogram analysis, spectrogram processing, and FFT-based methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pitch selection is performed using conventional methods, then pitch candidates are generated, but accuracy in selecting the correct pitch is reduced due to lack of local continuity and octave consistency checks
Solution Approach 1:
The pitch selection process is divided into separate functional modules: pitch candidate generation, local continuity checking, and regional octave consistency verification. Each module handles a specific aspect of pitch analysis, improving overall accuracy while maintaining manageable complexity through modular design.
Solution Approach 2:
The system performs preliminary pitch candidate generation and filtering before final pitch selection. By pre-processing pitch candidates through continuity and consistency checks, the system narrows down options early in the process, improving the accuracy of the final selection without excessive computational burden.
2Measurement precision
If buzz removal is not applied, then processing is simpler, but pitch detection accuracy deteriorates due to interference from buzz components
Solution Approach 1:
The system extracts and removes buzz components from the audio signal before pitch detection. By separating the harmful buzz elements from the target pitch signal, the system improves pitch detection accuracy while using dedicated removal modules that manage the added processing complexity.
Solution Approach 2:
The system identifies buzz components as potentially useful information for understanding signal characteristics, then selectively removes only the harmful portions. This approach converts the initially problematic buzz into an opportunity for refined pitch detection by understanding what needs to be removed.
3Measurement precision
If vibrato is not suppressed or flagged, then processing is simpler, but pitch selection accuracy deteriorates due to confusion between vibrato ornamentation and pitch alternation
Solution Approach 1:
The system performs preliminary vibrato detection and flagging before final pitch selection. By identifying vibrato patterns early in the processing chain, the system can distinguish between intentional ornamentation and actual pitch changes, improving accuracy in the subsequent pitch selection stage.
Solution Approach 2:
The vibrato detection module acts as an intermediary between raw pitch candidates and final pitch selection. It provides intermediate information about vibrato presence and characteristics, allowing the pitch selection algorithm to make more informed decisions without directly handling the complexity of vibrato analysis.
4Productivity
If multiple pitch candidates are considered without continuity checks, then pitch selection is faster, but reliability deteriorates due to lack of local consistency verification
Solution Approach 1:
The system applies partial continuity checking only to the most promising pitch candidates rather than performing exhaustive verification on all candidates. This selective approach maintains fast processing speed while providing sufficient reliability checks for the most likely correct pitches.
Solution Approach 2:
The system maintains continuous pitch tracking across adjacent frames, using the pitch selection from previous frames to inform current frame selections. This continuity ensures reliable pitch following through musical passages while keeping processing efficient by leveraging temporal consistency.
Data Source
AI summary
The technology disclosed relates to audio signal processing. It includes a series of modules that individually are useful to solve audio signal processing problems. Among the problems addressed are buzz removal, selecting a pitch candidate among pitch candidates based on local continuity of pitch and regional octave consistency, making small adjustments in pitch, ensuring that a selected pitch is consistent with harmonic peaks, determining whether a given frame or region of frames includes harmonic, voiced signal, extracting harmonics from voice signals and detecting vibrato. One environment in which these modules are useful is transcribing singing or humming into a symbolic melody. Another environment that would usefully employ some of these modules is speech processing. Some of the modules, such as buzz removal, are useful in many other environments as well.