Comparison of different implementations of MFCC

By Thomas Fang Zheng, Guoliang Zhang, Zhanjiang Song. Journal of Computational Science and Technology, 2001

640 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Experimental and Cognitive Psychology, Signal Processing

View Paper: https://www.semanticscholar.org/paper/6f145968059d006235d59d1c201d5185b440d221

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Experimental and Cognitive Psychology

In Signal Processing

Citations in the corpus

DeepEar: robust smartphone audio sensing in unconstrained acoustic environments using deep learning (2015, methodology): “Generally banks of either Perceptual Linear Prediction (PLP) [38] or Mel Frequency Cepstral Coefﬁcient (MFCC) [28] features are used as they are fairly effective across many audio modeling scenarios.”
Accelerating Mobile Audio Sensing Algorithms through On-Chip GPU Offloading (2017, methodology): “Application Purpose Main Features Inference Model Frame Window Emotion Recognition∗ [52] emotion recognition PLP∗ [32] 14 Gaussian Mixture Models∗ [18] 30ms 5s Speaker Identification∗ [52] speaker identification PLP∗ 23 Gaussian Mixture Models∗ 30ms 5s Keyword Spotting∗ [22] hotphrase recognition Fi”
Recent Advances in Google Real-Time HMM-Driven Unit Selection Synthesizer (2016, methodology): “It employs HMM emission probabilities trained using the ML criterion as target costs and the mel-frequency cepstral coefficient (MFCC) based ([6]) spectral differences for the join costs.”
A novel approach for MFCC feature extraction (2010, methodology): “Classification & Verification stage The GMM with expectation maximization is a feature modelling and classification algorithm widely used in the speech based pattern recognition, since it can smoothly approximate a wide variety of density distributions [17], [5].”
HiFiSinger: Towards High-Fidelity Neural Singing Voice Synthesis (2020, methodology): “However, they either leverage coarse-grained MFCC (Zheng et al., 2001) as acoustic features in slow autoregressive neural vocoder (Oord et al.”
An efficient MFCC extraction method in speech recognition (2006, background): “A filter bank is acceptable for speech recognition so far as its composite frequency response is flat over the entire frequency range of interest [10, 11].”
SoundSense: scalable sound sensing for people-centric applications on mobile phones (2009, methodology): “In this stage, we use MFCC as feature, which is designed to mimic human perception [ 45 ].”
Measuring the perceptual availability of phonological features during language acquisition using unsupervised binary stochastic autoencoders (2019, background): “Nonetheless, MFCCs are known to produce high-quality supervised speech recognizers (Zheng et al., 2001; Hinton et al., 2012), and we therefore leave optimization of the representation of speech features to future work.”