Audio Multimedia Conferencing, Voice Recognition-based Recording System

As a new communication method, multimedia conference can offer active situation of attendees in different meeting places, the content of the meeting , all types of data and information with the video and audio form to the attendees in each meeting place timely. So it became a quick, effective, economic and convenient new business which has a wide application. However, compared with the traditional meeting, current multimedia conference cannot record the speaker抯 statement in real- time. In this thesis, the technology of SR (speech recognition) is introduced into the audio multimedia conference system which is operated on a LAN. With the help of SR, the speech of any attendee is identified and translated from phonic signal into text signal that can be saved and edited later. This system has three logical parts: manage-node part, record-node part and conference communication part. Manage-node acts as a chairman to preside the conference and controls its progression when the conference is carrying on, while the conference is over it can provide the function of querying the former conference, editing the content of its record and informing relative departments of the coming conference, ect. Record-node takes effect only when the conference is carrying on. At this time, the record-node, which is appointed as the speaker, applying the technology of SR, translates the phonic signal into text signal that will be transmitted to the manage-node. Simultaneously it transmits the phonic signal to other record-nodes to make their users be able to hear the content of speech. Conference communication part transmits all the instruction and other altemant information among all the nodes to coordinate different functions and ensure the system to run smoothly.