My speech recognition is really low. What can I do?
The “Speech-to-Noise-Ratio” is used in speech recognition as a way of rating the quality of your microphone. A high ratio indicates that your speech, when compared to both sound (that occurs when you speak) and electronic noise (anything from computers to car engines that use or produce electricity), is quite strong relative to the noise signals. A low ratio indicates that you may be speaking too loudly or even too quietly into the mask. Refer to the “Proper Breathing Technique” section in the manual that you received with your Stenomask for proper breathing and speaking Techniques. It takes practice to get good recognition but the system does work very well once you master the techniques!
In DNSv14 this ratio number disappeared, for a time there was a level number, and on and off there has been a green bar meter.
The optimal settings for the highest performance in Dragon Naturally Speaking or Dragon Professional are:
A sound card capable of supporting 22 kHz 16-bit stereo audio recording.
Above this recording level you increases the amount of data and processing that the speech engine will have to sort through and slows down your dropdown and potentially your system as well.
Using a single channel adapter or mono adapter and microphone will reduce accuracy by a small amount since there is no differential comparison in the channel signals. Using a higher bit rate will still result in the above increased overhead and processing as each channel is handled in its own thread, so the amount of data per channel is the same and causes an increased load on that processing thread.
Using an primary SSD hard drive, and NVME, and if it is an option a secondary SSD or HDD will allow the fastest responses and dropdown from Dragon. The OS and Application would run from the NVME (the fastest drive), data (dra files etc) get configured to be written to the secondary SSD or HDD so that data bandwidth is not reduced between the application and the data storage, this becomes especially important when Dragon starts to append and concatenate its dra file over longer sessions.
A more recent issues is the shared access to the soundcard, Dragon needs to be able to control the gain level to move/keep the audio signal in the center of the channel to avoid clipping and clamping of the signal (cutting off the top and bottom parts of the signal). As long as Dragon has exclusive control, shared access should be okay.
Reported problem with Play 3: Default impedance setting is incorrect and does not function correctly.
Reported problem with Play 4: Firmware has to be forcibly upgraded to function correctly.
