Search

15 Aug 2007: This increases the likelihoodcalculation cost from O(n) to O(n2) for an n-dimensional featurevector. ... Thusthe cost for accumulating statistics is simply O(n3) for all n dimen-sions.

19 Jun 2007: converges with very few iterations actually requiring a summation, which is a significant cost interms of processing time. ... Lateral and elevational tracking can be included at a small additionalprocessing cost and enable strain calculations over a

10 Oct 2007: However, in order totrain the acoustic models word level transcriptions are also required.Hence the major cost of using large amounts of broadcast data is theprovision of accurate manual transcriptions. ... It can be seen that the reductions in CER

3 May 2007: Furthermore,it may result in large memory and runtime costs for the system. ... 4.3.3 Memory and Computational Cost. Two important issues in speech recognition (and many other applications) are the memory andruntime computational cost of the system.

25 May 2007: It bears noting that the main computational expense of normalisation comes from fitting aparametric displacement or strain surface, but this in itself is typically a negligible cost on widelyavailable GHz processors ... The normalisation surface might

22 Jun 2007: Email: {lw256, mjfg, pcw}@eng.cam.ac.uk. ABSTRACT. A significant cost in obtaining acoustic training data is the genera-tion of accurate transcriptions.

22 Jun 2007: However, sinceit may break a possible phrase that could be translated in its entirety,the cost in terms of translation score may be far higher.

4 Apr 2007: For pMPE there is a slight increase in this cost. Thelikelihood of the model parameters, θ = {µsmj , σ2smj}, given the observationvector, ot, is given by. ... 2 The additional cost is due to the computation of the posterior probabilities forthe

30 May 2007: Note in recent years MBR decoding,associated normally with the word-level cost function, hasbecome popular in speech recognition [24], [25], [26]. ... A log-posterior cost function is used, as in theMMI criterion, rather than the posterior-based MCE and

23 Apr 2007: As it is hard to efficientlycontrol the computational cost, this approach is only applicableto systems with small number of adaptation parameters, forexample, cluster adaptive training [4]. ... The computational cost thenincreases exponentially.2.