A dialogue system should capture speakers' intentions, which can be represented by combinations of speech acts, predicators, and sentiments.
The new events pages can also capture speakers, exhibitors, sponsors and awardees.
Unlike these approaches, which is based on a joint model, our method trains two exclusive RBMs for each speaker, aiming to capture speaker-specific conversion-friendly features.
The quality and naturalness in the VC system may be improved by capturing speaker-specific features in the high-frequency region.
Using these matrices, an input vector x (t) at frame t given the previous vector x (t−1) is projected into the speaker-dependent latent space that captures speaker-individualities.
In addition to these methods, other models are applied to capture speakers, including eigenvectors to model and adapt speakers [13], nonnegative matrix factorization-based models [14, 15], and sinusoidal models [16].
It was verified that the frames selected by the proposed method captured more speaker information than a VAD.
Beyond the traditional GMM-based LVCSR techniques, appending identity vectors (i-vectors), which capture speaker and channel information in a low-dimensional representation, to each frame of DNN input has become a promising approach for making DNN more speaker invariant [8, 9].
The i-supervector extractor can be implemented by adopting the diagonal modeling part of the JFA [12],[24] with a slight modification: the diagonal model D is trained per utterance instead of per speaker basis in order to capture both speaker and session variability.
Rather than making a distinction between the speaker and channel effects, the total variability space in the i-vector extraction method simultaneously captures the speaker and channel variabilities [27, 28].
