Abstract: In this study, a system that captures the spirit of sound detective work is presented. Similar to a musician, it analyzes the sound by first removing certain elements, recognizing individual instruments, confirming the notes or sounds they produce. The finished product is like to a information of how many instruments are involved in a particular audio file. It can be a fresh score or text, with the detection of pitch feature. It is available for listening, and people can enjoy the music immensely. This project accomplishes a number of things really effectively for listening to polyphonic music performed by multiple instruments at once. The implementation of automated project that allows the users to identify the instruments involved in the audio files and detect the pitches out of the detected instruments. We first dealed with the analyzing of monophonic instruments sound which involves only one instrument and got the accuracy of 70 % for 40 Epoch. While dealing with polyphonic instrument sound there were many complexity faced during the dataset collection and testing phase. From previous survey there is work only done on monophonic instruments .Since there were no work done on polyphonic instruments sound, there was no dataset available. The merging of the monophonic instrument sound taken from different websites were done, almost 30,000 audios were collected .The one audio file contains varying number of instruments merged .Some contains four ,some five instruments. For training it is 2630 and testing 81 audio files. While the user can provide any of audio files from total of 2462 audio files. The accuracy decreased since we are dealing with very complex, overlapping datasets. Re-labelling would provide good results. We got 30% accuracy with 40 Epoch. The detection of Instruments was determined. The Other features includes pitch detection and synchronization. The real-time pitch detection using the Web Audio API and microphone input, along with the ability to analyze pre-recorded audio files is implemented. The synchronization process allows the user to upload an audio file, analyze it, generate synchronized MIDI data, and present the synchronized output back to the user for further interaction or analysis. The goal is to enhance user satisfaction by providing music that aligns with their instrument detection, pitch detection and synchronization.
Keywords: Instrument Detection, Pitch Detection, Synchronization, MIDI Synthesizer, Polyphonic, Monophonic.
|
DOI:
10.17148/IJARCCE.2024.13504
