Feature warping for robust speaker verification
By Jason W. Pelecanos, Sridha Sridharan. The Speaker and Language Recognition Workshop, 2001
784 citations recorded by the source corpus.
View Paper: https://www.semanticscholar.org/paper/f39d0c70e0f89296045f0f11d79e15fa978e3081
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
Citations in the corpus
- End-to-End attention based text-dependent speaker verification (2016, methodology): “A rolling-window based cepstral mean normalization (CMN) and feature warping [35] are applied in these systems.”
- Front-End Factor Analysis for Speaker Verification (2011, methodology): “This 20-dimensional feature vector was subjected to feature warping [15] using a 3, s sliding window.”
- Advances in deep neural network approaches to speaker recognition (2015, background): “The decision tree is grown by asking a set of locally optimal questions that give the largest likelihood increase, assuming that the data on each side of the split can be modeled by a single Gaussian.”
- A review of depression and suicide risk assessment using speech analysis (2015, background): “(2011) indicate that feature warping, as a per-speaker feature space normalisation technique (Pelecanos and Sridharan, 2001), offers little to no improvement in a presence of depression classifier.”
- Joint Factor Analysis Versus Eigenchannels in Speaker Recognition (2007, background): “…term in the right-hand side of (3) is modeled by assuming that if is the speaker supervector for a randomly chosen speaker then (5) where is the speaker- and channel-independent supervector, is a rectangular matrix of low rank, is a diagonal matrix, and and are random vectors having standard…”
- An overview of text-independent speaker recognition: From features to supervectors (2010, background): “Feature warping (Pelecanos and Sridharan, 2001) and short-term Gaussianization (Xiang et al.”
- A Study of Interspeaker Variability in Speaker Verification (2008, methodology): “These were subjected to feature warping using a 3-s sliding window [22]. coefficients were then calculated using a five-frame window, giving a total of 40 features.”
- LIUM SPKDIARIZATION: AN OPEN SOURCE TOOLKIT FOR DIARIZATION (2010, methodology): “The entire features of the recording are warped using a 3 second sliding window as proposed in [10], before the features of each cluster are normalized (centered and reduced).”
- Support vector machines versus fast scoring in the low-dimensional total variability space for speaker verification (2009, methodology): “This 20 -dimensional feature vector was subjected to feature warping [8] using a 3 s sliding window.”
- Comparison of Speaker Recognition Approaches for Real Applications (2011, unknown): “Feature warping to a Gaussian distribution is performed on a 3 sec sliding window excluding silence frames [10].”
- Speaker recognition by means of Deep Belief Networks (2013, methodology): “These features were computed with a frame rate of 100 observation vectors per second, and were subject to short term gaussianization [20] computed on a 3 sec sliding window applied to speech frames only.”
- An overview of automatic speaker diarization systems (2006, methodology): “Different types of feature normalization have been used with this process, namely RASTA-filtered cepstra with 10-s feature mean and variance normalization [15] and feature warping [44] using a sliding window of 3 s [14], [17].”