Applies to older version SpeechServers (Dragon v5/6 or IBM v8+) only. Older version included Standalone desktop client edition that is no longer available.
New user profiling or supplementary general training is provided by CompleteProfile. Profiling is handled through the automatic activity of CSUSA_Profile_X where X is the engine type. The activity must exist on a workflow for processing.
Profiling will use an audio file and a verbatim text file. This will provide initial "setup" when creating a new user in an automated fashion.
The source audio file can be in any supported acWAVE format (.wav, .vox, .mp3, .wma) and will be converted to the engine's requirements before profiling. To insure maximum accuracy, use a high-quality recording (low bit rates can cause more inaccuracies).
For the Dragon version, an audio file of at least 15 minutes of continuous audio is required for proper profiling. The source verbatim text file must be in a .txt format.
For the IBM version, at least a 1,000 words must be correctly transcribed for the profiling to be successful. If one profile has been completed, the requirement increases to 2,000, then 3,000, and so on. This means that the engine must transcribe a 2,000 word text with at least 50% accuracy to begin the process. Alternatively, a 10,000 word text with 10% accuracy or 5,000 word text with 20% accuracy could be used. During the second round, the engine must transcribe, for example, a 4,000 word text with 50% accuracy. The source verbatim text must be in .txt format.
If the job does not have an assigned engine user ID (based on Author or Job UDA settings), the engine will automatically create a new user (based on default settings) for processing. This new user will be assigned to the job's author.
When profiling starts, the user is locked in the repository, preventing other servers from updating the same user. Once complete, the user is uploaded to the repository and released for further use. Once uploaded, all other servers will use the newly profiled user for future activities.
The activity will complete normally if no errors occur, or will complete with a condition of ERROR in the event that an error occurred.
Both the Dragon and IBM version may be used to further train the system after an enrollment is created interactively by the speaker in front of a PC. The user's post-enrollment speech files may be transferred to the server for use in the User Repository, this will allow the training of a user on the server with training files already created on the local machine. This traditional enrollment can be further trained using CompleteProfile and/or SpeechTrainer.
To perform training, do the following:
1. Create a job in Command! on a workflow that has the CSUSA_Profile_X activity and assign (or create) the required audio file. Also assign (or create) the verbatim text file.
2. Be sure to have appropriate UDA settings according to SpeechParameters.