Abstract: A voice processing system comprising a digital trunk processor 135 and a system unit 145 incorporates a local voice recognition resource 185 provided on a digital signal processing board 180, and can access a remote voice recognition resource 320 in a server 300 via a local area network 250. The system provides a barge-in facility for the recognition resource on the remote server, whereby a prompt is played out to the user, and the incoming telephony signal is fed into a voice activity detector on the digital trunk processor. Responsive to a detection of incoming voice activity, the outgoing prompt is terminated, and the incoming data is transferred over the local area network to the remote server for recognition.

Abstract: A voice processing system comprising a digital trunk processor 135 and a system unit 145 incorporates a local voice recognition resource 185 provided on a digital signal processing board 180, and can access a remote voice recognition resource 320 in a server 300 via a local area network 250. The system provides a barge-in facility for the recognition resource on the remote server, whereby a prompt is played out to the user, and the incoming telephony signal is fed into a voice activity detector on the digital trunk processor. Responsive to a detection of incoming voice activity, the outgoing prompt is terminated, and the incoming data is transferred over the local area network to the remote server for recognition.

Abstract: A method and system for processing a telephone call in a distributed system comprising a voice response unit and a resource server connected by a local area network; wherein the voice response unit utilizes one or more digital trunks, each of the trunks having a plurality of telephone lines, and wherein the resource server includes a voice recognition system. The method and system comprise demultiplexing an incoming signal, aggregating the incoming telephone signal into a succession of voice data packets, transmitting the packets over the local area network, performing a voice recognition function on the succession of packets, and transmitting the recognized text over the local area network.

Abstract: A speech recognition system for discrete words uses a single Hidden Markov Model (HMM), which is nominally adapted to recognise N different isolated words, but which is trained to recognise M different words, where M>N. This is achieved by providing M sets of audio recordings, each set comprising multiple recordings of a respective one of said M words being spoken. Only N different labels are assigned to the M sets of audio recordings, so that at least one of the N labels has two or more sets of audio recordings assigned thereto. These two or more sets of audio recordings correspond to phonetically dissimilar words. The HMM is then trained by inputting each set of audio recordings and its assigned label. The HMM can effectively compensate for the phonetic variations between the different words assigned the same label, thereby avoiding the need to utilise a larger model (i.e., to use M labels).

Abstract: A first computer workstation 10 includes a voice response unit for interfacing to a telephone network. The first computer workstation is attached by a communications link 18 to a second computer workstation 20, which includes a server to perform a particular voice processing function, such as text to speech, voice recognition, FAX-back, and so on. For inbound applications the first computer workstation forwards the incoming signal over the communications link to the server on the second computer workstation for real-time processing, whilst for outbound applications, the reverse process occurs.