The fine spectral structure related to pitch information is conveyed in Mel cepstral features, with variations in pitch causing variations in the features. For speaker recognition systems, this phenomenon, known as “pitch mismatch” between training and testing, can increase error rates. Likewise, pitch-related variability may potentially increase error rates in speech recognition systems for languages such as English in which pitch does not carry phonetic information. In addition, for both speech recognition and speaker recognition systems, the parsing of the raw speech signal into frames is traditionally performed using a constant frame size and a constant frame offset, without aligning the frames to the natural pitch cycles. As a result the power spectral estimation that is done as part of the Mel cepstral computation may include artifacts. Pitch synchronous methods have addressed this problem in the past, at the expense of adding some complexity by using a variable frame size and/or offset. This paper introduces Pseudo Pitch Synchronous (PPS) signal processing procedures that attempt to align each individual frame to its natural cycle and avoid truncation of pitch cycles while still using constant frame size and frame offset, in an effort to address the above problems. Text independent speaker recognition experiments performed on NIST speaker recognition tasks demonstrate a performance improvement when the scores produced by systems using PPS are fused with traditional speaker recognition scores. In addition, a better distribution of errors across trials may be obtained for similar error rates, and some insight regarding of role of the fundamental frequency in speaker recognition is revealed. Speech recognition experiments run on the Aurora-2 noisy digits task also show improved robustness and better accuracy for extremely low signal-to-noise ratio (SNR) data.

The relative importance of the temporal characteristics of speech for text-dependent and text-constrained speaker verification is investigated. A novel scheme is proposed using a common set of Gaussian components to form various HMM and GMM configurations, establishing a systematic transition from text-dependent to text-constrained speaker verification, and resulting in a novel alternative to conventional GMM-UBM training. Experimental results indicate that the intra-word temporal characteristics of speech do not contribute significantly to performance, however the inter-word temporal characteristics can be used during both enrollment and testing to improve verification performance.

In large-scale deployments of speaker recognition systems the potential for legacy problems increases as the evolving technology may require configuration changes in the system thus invalidating already existing user voice accounts. Unless the entire database of original speech waveform were stored, users need to reenroll to keep their accounts functional, which, however, may be expensive and commercially not acceptable. We define model migration as a conversion of obsolete models to new-configuration models without additional data and waveform requirements and investigate ways to achieve such a migration with minimum loss of system accuracy. As a proof-of-concept, an algorithm for statistical migration in the Maximum A-Posteriori framework is studied and evaluated experimentally using the NIST SRE-03 dataset. The migration step is discussed in a wider conceptual framework of Conversational Biometrics.

This paper presents an overview of the architecture and algorithms implemented in IBM's text-independent speaker verification system developed for the 2002 NIST speaker recognition evaluation, particularly for the 1-speaker detection task using cellular test data. We describe individual components including a Gaussianization front-end, celluar-codec post-processing, modeling, discriminative optimization and scoring steps. A combination of multiple, data-perturbed systems using a discriminative objective so as to achieve optimum performance for a low false alarm operating region obtained the top performance in the NIST 2002 1-speaker detection task.

Pitch information is known to be partially conveyed in Mel cepstral features that are commonly used for speaker recognition. In particular, for high pitched female speakers, and whenever average pitch varies significantly between enrollment and testing, the fine spectral structure introduced by the fundamental frequency was shown to degrade speaker recognition performance. This paper introduces a signal processing procedure termed depitch that attempts to remove pitch information from the speech signal. Recognition experiments carried out on the female subset of the NIST 2002 Speaker Recognition Evaluation show that by combining scores from a conventional and a depitched system, a substantial improvement in equal error rate is obtained for high pitched speakers and pitch-mismatched trials. Performing pitch/depitch score fusion is also shown to help alleviate the well-known problem of "goat" speakers.

Conversational Biometrics combines acoustic speaker verification with conversational knowledge verification to make a more accurate identity decision. To manage the added level of complexity that the multi-modal user recognition approach introduces, this paper proposes the use of verification policies,in the form of Finite State Machines, which can be used to program a policy manager. Once a verification policy is written, the policy manager interprets the policy on the fly, and at every turn in the session decides dynamically whether to accept the user, reject the user, or continue to interact and collect more data. The policy manager allows for any number of verification engines to be plugged-in, thereby adding to the flexibility of the framework. As a result, system developers have significant freedom to design user verification solutions, and a wide variety of applicationspecific, transaction-specific and user-specific constraints can be addressed using a generic system. The paper also describes a prototype implementation of a Conversational Biometrics solution with the proposed programmable policy manager.

Pitch mismatch between enrollment and testing is a common problem in speaker recognition systems. It is well known that the fine spectral structure related to fundamental frequency manifests itself in Mel cepstral features used for speaker recognition. Therefore pitch variations result in variation of the acoustic features, and potentially an increase in error rate. A previous study introduced a signal processing procedure termed depitch that attempts to remove pitch information from the speech signal by forcing every speech frame to be pitch synchronous and include a single pitch cycle. This paper presents a modification of the depitch algorithm, termed syncpitch, that performs pseudo pitch synchronous processing while still preserving the pitch information. The new algorithm has a relatively moderate effect on the speech signal. System combination of syncpitch with a baseline system is shown to improve speaker verification accuracy in experiments conducted on the 2002 NIST Speaker Recognition Evaluation data.

This paper provides a description of the properties of the symmetric sphericity measure between two Gaussian classes, and presents experimental results of its use in the context of text independent speaker verification, performed on cellular speech. A novel geometric interpretation of the sphericity measure is presented, emphasizing its robustness to variations in scaling in feature space compared to other distortion measures, along with an associated score normalization procedure. The experimental results clearly indicate the superiority of this method over the prevailing likelihood ratio approach, motivating further use of the sphericity measure in a multi-component modeling scheme. In addition, the direct use of this method with single Gaussian models is computationally extremely simple, and may provide acceptable performance in certain cases.

This paper describes a computationally simple method to perform text independent speaker verification using second order statistics. The suggested method, called utterance level scoring (ULS), allows one to obtain a normalized score using a single pass through the frames of the tested utterance. The utterance sample covariance is first calculated and then compared to the speaker covariance using a distortion measure. Subsequently, a distortion measure between the utterance covariance and the sample covariance of data taken from different speakers is used to normalize the score. Experimental results from the 2000 NIST speaker recognition evaluation are presented for ULS, used with different distortion measures, and for a Gaussian mixture model (GMM) system. The results indicate that ULS as a viable alternative to GMM whenever the computational complexity and verification accuracy needs to be traded.

This letter describes speaker verification using a covariance modeling approach for speaker and world modeling. Two verification methods are suggested: frame level scoring and utterance level scoring. Both methods exhibit extremely low computational and model-storage requirements. The suggested methods are tested on the male segment of the 1999 NIST Speaker Recognition Evaluation corpus, using a single training session, and compared to a Gaussian mixture model (GMM) system. The degradation in accuracy and the computational requirements are estimated. Covariance modeling is seen to be a viable alternative to GMM whenever computational and storage requirements must to be traded with verification accuracy.

We are interested in automatically detecting specific phone segments that have been mispronounced by a nonnative student of a foreign language. The phone-level information allows a language instruction system to provide the student with feedback about specific pronunciation mistakes. Two approaches were evaluated; in the first approach, log-posterior probability-based scores [1] are computed for each phone segment. These probabilities are based on acoustic models of native speech. The second approach uses a phonetically labeled nonnative speech database to train two different acoustic models for each phone: one model is trained with the acceptable, or correct native-like pronunciations, while the other model is trained with the incorrect, strongly nonnative pronunciations. For each phone segment, a log-likelihood ratio score is computed using the incorrect and correct pronunciation models. Either type of score is compared with a phone dependent threshold to detect a mispronunciation. Performance of both approaches was evaluated in a phonetically transcribed database of 130,000 phones uttered in continuous speech sentences by 206 nonnative speakers.