Feature Extraction Using Mfcc

Gupta, Shikha; Jaafar, Jafreezal; Fatimah, Wan; Bansal, Arpit

doi:10.5121/sipij.2013.4408

Cited by 89 publications

(37 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Frame blocking bertujuan untuk memecah sinyal suara menjadi blok frame kecil. Frame tadi berisi sampel yang cukup untuk mendapatkan informasi [2]. Sedangkan windowingberfungsi untuk mengurangikesenjangandi awal dan akhir dari frame windowing [6].…”

Section: B Preemphasis Frame Blocking Windowing Dan Fftunclassified

Modifikasi Metode MFCC untuk Identifikasi Pembicara di Lingkungan Ber-Noise

Prayogi

2019

JOINTECS

View full text Add to dashboard Cite

Some feature extraction methods for speaker identification systems have a disadvantage that their accuracy decreases when the environment is noisy.The Mel Frequency Cepstral Coefficient (MFCC) method is a noisesensitive voice extraction method.The MFCC method produces high accuracy when in a clean environment. Conversely, when in a noise environment, accuracy drops dramatically.This research proposes feature extraction methods using MFCC combined with endpoint detection algorithm.The endpoint detection algorithm separates speech and nonspeech regions.Nonspeech area usually contains more noise so it can be used as noise information.The noise information is extracted and generates a noise frequency magnitude.Test of the proposed method yields higher accuracy values for all noise types and SNR levels.The accuracy produced by the proposed method is 14.69% higher than the MFCC method, 6.4% compared to the MFCC + wiener method, and 2.74% compared to the MFCC + Spectral Subtraction (SS) method.

show abstract

Section: B Preemphasis Frame Blocking Windowing Dan Fftunclassified

Modifikasi Metode MFCC untuk Identifikasi Pembicara di Lingkungan Ber-Noise

Prayogi

2019

JOINTECS

View full text Add to dashboard Cite

show abstract

“…The last step is to compute Discrete Cosine Transform (DCT) of the log filterbank energies. A DCT is a fourier related transform similar to the Discrete Fourier Transform (DFT), but using only real numbers [14]. It has a property of energy packing or compaction.…”

Section: Mel Frequency Cepstral Coefficientsmentioning

confidence: 99%

“…Also, it decorrelates the overlapping filterbank energies. First 13 coefficients are retained because the increasing number of coefficients represent faster change in the estimated energies and thus have less information for classifying audio signals [14]. …”

Section: Mel Frequency Cepstral Coefficientsmentioning

confidence: 99%

Comparison of time series similarity measures for plagiarism detection in music

Suneja

Bansal

2015

2015 Annual IEEE India Conference (INDICON)

View full text Add to dashboard Cite

Music is present everywhere around us. It is present in car rides, hotels, homes, television shows, movies, etc. With a huge demand of songs for bands, movies, etc., writers and singers are pressurised to produce new songs, but face the challenge of ensuring that they are not copying an already existing song in any way. With the growing music industry, cases of plagiarism have become a critical concern for musicians. An enormous number of musical tracks are released every year. So, there arises a need of a reliable and easy way to search through the huge database of songs that match the query song. In this paper, we have implemented the five similarity measure algorithms in MATLAB and did the comparative analysis of using them to distinguish among three sets of songs: a pair of plagiarized songs, a pair of same song with different lengths, and a random pair of songs to find the best fit algorithm for plagiarism detection.

show abstract

“…The only helpful data is selected in feature extraction that describes the signal and undesirable data is excluded. MFCC is a famed and excellent method for feature extraction from a speech signal [13], [14] that can be also used for face, gesture, palm print, satellite and iris image identification [15]- [19].…”

Section: Recognition Systemmentioning

confidence: 99%