As Voice over IP proliferates, Packet Loss Concealment (PLC) at a receiver has emerged as an important factor in determining the Quality of Service. Through the heuristic use of signal repetition and overlap-add interpolation to handle packet loss, conventional PLC systems largely ignore the dynamics of the statistical evolution of the speech signal, often leading to perceptually annoying artifacts. To address this problem, we propose the use of statistical models for PLC in which long-term signal evolution is exploited. In particular, a Hidden Markov Model (HMM) is used to model and track the evolution of speech signal parameters such as spectral envelope, pitch, voicing, and energy. Within the HMM framework, a Minimum-Mean-Squared Error (MMSE) approach is utilized to provide estimates of the missing speech parameters for both the extrapolation and interpolation procedures of PLC. Our studies show that the HMM-based PLC provides more accurate speech signal parameter substitution than conventional PLC systems. Therefore, the HMM-based PLC produces speech parameters that more accurately reflect the context of the speech signal evolution in contrast to conventional methods [C.A. Rødbro, M.N. Murthi, S.V. Andersen and S.H. Jensen 2004]. (Christoffer A. Rødbro, Manohar N. Murthi (University of Miami), Søren Vang Andersen, Søren Holdt Jensen)
|Effective start/end date||19/05/2010 → 31/12/2017|
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.