Dynamic behaviour of connectionist speech recognition with strong latency constraints [Elektronisk resurs]

• This paper describes the use of connectionist techniques in phonetic speech recognition with strong latency constraints. The constraints are imposed by the task of deriving the lip movements of a synthetic face in real time from the speech signal, by feeding the phonetic string into an articulatory synthesiser. Particular attention has been paid to analysing the interaction between the time evolution model learnt by the multi-layer perceptrons and the transition model imposed by the Viterbi decoder, in different latency conditions. Two experiments were conducted in which the time dependencies in the language model (LM) were controlled by a parameter. The results show a strong interaction between the three factors involved, namely the neural network topology, the length of time dependencies in the LM and the decoder latency. 

Ämnesord

Engineering and Technology  (hsv)

Mechanical Engineering  (hsv)

Fluid Mechanics and Acoustics  (hsv)

Teknik och teknologier  (hsv)

Maskinteknik  (hsv)

Strömningsmekanik och akustik  (hsv)

Natural Sciences  (hsv)

Computer and Information Sciences  (hsv)

Computer Sciences  (hsv)

Naturvetenskap  (hsv)

Data- och informationsvetenskap  (hsv)

Datavetenskap (datalogi)  (hsv)

Humanities and the Arts  (hsv)

Languages and Literature  (hsv)

Specific Languages  (hsv)

Humaniora och konst  (hsv)

Språk och litteratur  (hsv)

Studier av enskilda språk  (hsv)

TECHNOLOGY  (svep)

Engineering physics  (svep)

Acoustics  (svep)

TEKNIKVETENSKAP  (svep)

Teknisk fysik  (svep)

Akustik  (svep)

TECHNOLOGY  (svep)

Information technology  (svep)

Computer science  (svep)

TEKNIKVETENSKAP  (svep)

Informationsteknik  (svep)

Datavetenskap  (svep)

HUMANITIES and RELIGION  (svep)

Languages and linguistics  (svep)

HUMANIORA och RELIGIONSVETENSKAP  (svep)

Språkvetenskap  (svep)

Indexterm och SAB-rubrik

speech recognition; neural network; low latency; non-linear dynamics