Automatic speech recognition systems convert the speech in an audio signal into text. Best results are usually obtained by using an optimised profile for the particular person’s voice, typically achieved through the use of a “respeaking” technique (see below).
WINCAPS Integration with Speech Recognition
WINCAPS QLIVE and the PRO version of WINCAPS QU4NTUM both provide a light, integrated UI for Dragon NaturallySpeaking® Professional software (not included). This optimises the input from the speech recognition system for either live or offline transcription, so there’s no need to run the Dragon toolbar separately, giving a slightly faster response and using fewer system resources while ensuring that recognised text is always directed where it is required (in the subtitle document) regardless of current cursor position. WINCAPS adds further value by enforcing a house style, interpreting spoken style control commands, smoothing the delivery rate and providing for keyboard intervention/correction if required.
- Maintains text input path regardless of application focus
- Corrects consistently misrecognised words or phrases
- Improved delivery time
- Reduced loading on system resource
- Keyboard intervention without pause in re-speaking
Dragon NaturallySpeaking® is the most popular speech recognition system in use for live subtitling due its speed and accuracy, however other real-time speech recognition engines may also be supported using a standard keyboard emulation interface rather than the dedicated SDK.
Dragon NaturallySpeaking® is a registered trademark of Nuance, Inc. and is used here under license.