Future work will examine other speech materials produced with more natural tempo variations.
For parameter training of parametric shaping functions, we used a number of speech materials from the TIMIT database, as will be described further in Section 4.
For speech materials, we randomly selected total 30 files spoken by four male and female native Korean speakers from the NTT Korean speech database where each phrase consists of two different meaningful sentences and the duration of each sentence was 8 s.
In (8), to find the optimal parameters, we compute the average of MSEs over all training data in the speech materials by varying α, β, and γ: begin{array}rcl@ ({alpha^ ast}, {beta^ ast}, {gamma^ ast})=argmin_{alpha, beta, gamma} xi alpha, beta, gamma) end{array} (9).
To find the optimal parameters, we compute the average of F in each objective measure over all training data in the speech materials by varying α, β, and γ and taking the optimal parameters α ∗, β ∗, and γ ∗ to be those that satisfy the below: begin{array}rcl@ ({alpha}^{ast}, {beta}^{ast}, {gamma}^{ast})=argmax_{alpha, beta, gamma} xi alpha, beta, gamma).
Methods: Taking into account the rules for developing speech materials and the unique linguistic characteristics of Mandarin, we designed and digitally recorded a set of materials comprised of seven lists of monosyllabic words, nine lists of disyllabic words, and fifteen lists of sentences with a high degree of subject familiarity.
The cabinet secretary insisted to MPs that his guidance to civil servants was not meant to deny their ministers "factual material" but only "briefing and speech material" supporting the leave campaign.
"Civil servants won't provide briefing and speech material for those who want to argue against the government's position, but in every other respect they will get the full service that you would expect.
The document said it would not be "appropriate or permissible" to provide briefing or speech material to ministers opposing the government's position, adding that "official departmental papers" were included in that.
These approaches require annotated language-dependent speech material in addition to annotated user-specific speech material.
