|
|
NESTOR
Distributed Speech Transcription System |
|
Nestor is a network based distributed transcription and documenting system, designed to make transcription faster, more efficient and of better quality. It provides automated work flow to multiple typists for fast documenting of huge volumes of speech.
Audio data acquired by recording stations and stored at a separate sound server, are subsequently automatically distributed between typists. Each of them then types the transcription of one short sound fragment. All fragments of sound and text are finally connected into a single document. The whole text including linked sound is available at the editor’s workstation to check, correct, print and archive.
APPLICATION Simultaneous transcription (composing verbatim protocol) of speeches, meetings, conferences hold by governmental and court bodies, public and commercial organizations.
BRIEF DESCRIPTION OF THE COMPLEX
Speech signal coming to the input of the system is recorded into the hard disc of PC. Then it is divided into fragments and distributed between operators-typists, who listen to and type. The text fragments are automatically jointed in the single document which can be saved and printed.
CASE STUDY
Complex “Nestor” is already installed and successfully working in:
- President administration of Russian Federation (RF)
- Council of Federation of RF Assembly
- Government of RF
- Constitutional Court of RF
- Ministry of Economy and Trade of RF
Others:
- Central Election Committee
- Supreme Court of Byelorussia
- Administrations of 21 regions of Russia
The complex consists of:
- System’s server
- Recording station
- Working stations “Record manager”
- Working station “Administrator”
- Working station “Editor”
- Working station “Operator”
Technical characteristics:
- Digital recording of speech sessions using 8 mono channels or 4 stereo channels with one audio card. The number of cards can be enlarged.
- Managing recording process and acoustic control of sound data.
- Distribution of sound fragments of speeches to working stations of editors and typists.
- Fast verbatim report preparation by means of parallel independent processing of small fragments by several operators and automatic “gluing” of the transcript fragments in a single final file.
- Using of MS Word for typing
- Text messenger between components of the complex
- Archiving of sound and text files
- Fast search of the text reports and the corresponding audio files.
- Sound recording from internal and external audio cards
- Reliable software and hardware reservation of complex’s subsystems.
Server
Server – is a key link of the system. All components of “Nestor” has connection only with the server of the system sending request and receiving results of their fulfillment. Software of the Server includes DBMS Microsoft SQL Server. Client’s working stations makes request to DBMS of the Server by means of OLEDB. All significant data which are important for the work of the Complex are saved on the Server.
- Sound files
- Information about the recordings
- Information about the projects
- Segmentation of the recordings on fragments
- Final reports of speeches
- Sound fragments
The second (reserve) server component makes its soft-and hardware reservation. Synchronization between the servers is built on the basis data replication by means of DBMS Microsoft SQL Server.
Record station
The record station makes input, analogue-digital processing of speeches and their record by 8 channels at the same time
Record station includes the following functional components:
- Input audio card H200, H205, H205L
- External input audio card H216
- Drivers of audio cards
- Subsystem of sound data reading
- Subsystem of reserve data storage
- Subsystem of data transferring
Record station is designed as a service with automatic restart which makes it stable to fails. It combines the functions of digital recorder and a network application. And it also includes the following possibilities:
- Synchronous record on the hard disc
- Transferring of the additional data on the server
- Network connection with the server
- Reserve copying of the received data
- Synchronous transferring of the data on the server
- Mono/stereo switching
Record manager
This is the interface for record station management realized with the usage of modern tools for GUI business application development
- Remote control of record station
- Several record stations support
- Spectrum analyzer for any channel
- Record start on the reserve channels
- Sound card adjustment and acoustic control on any channel
Administrator
Administrator’s working station is designed for management of protocol composing process.
Administrator can:
- Define the roles of the working stations in the complex
- Delete records and projects
- Create project and groups for their processing
- Make settings for the project
- Switch operators and editors between projects
- View the statistics of the operators’ and editors’ work for a day
- Control the degree of readiness of the final document and fragments
- Send informational messages to any of the components of the complex
- View the information on the complex
Editor
Editor’s working station is a module for MS Word. After being installed the module automatically launches every time Word is started.
Editor is intended to assemble the final document from the fragments of the record, processed by operators. Editor can also participate in the process of transcription.
Operator
Operator’s working station is intended to prepare fragments of the record. Operator’s working station is a module for MS Word. After being installed the module automatically launches every time Word is started.
Possible configurations
- 6 users (3 operators. 1 editor, 1 record manager, 1 administrator)
- 10 users (7 operators, 1 editor, 1 record manager, 1 administrator)
- 17 users (12 operators, 3 editor, 1 record manager, 1 administrator)
- 33 users (26 operators, 5 editor, 1 record manager, 1 administrator)
|
