- 3 Minutes to read
- 3 Minutes to read
How to Say...
About nVoq Dictation
nVoq HIPAA-compliant medical dictation transcribes the words you say. In some cases, nVoq formats the text based on common medical documentation standards and appropriate styles of grammar. If you prefer alternative formatting for specific text returns, please explore the use of Substitutions within our support documentation.
Best Practices for Better Dictation Accuracy
Whether you use an nVoq dictation client or a client of one of our partners, there are things you can do to improve dictation accuracy.
- Position your microphone correctly.
Don't place the microphone in the direct line of speech. Keep a headset microphone approximately 2-3 fingers below your mouth, and a handheld microphone should be held about chest high. If your microphone is too close the recording will include hisses and breathing noises, making it harder to provide an accurate transcript. Placing the microphone directly in front of your mouth will lead to "P-pops" (also called a "plosive" which happens when a blast of air hits the microphone). If the microphone is too far away, it will pick up background noise.
We recommend that you use a microphone that is supported by nVoq, but be sure to test any microphone you use to make sure you get acceptable dictation accuracy. For information about microphones that nVoq supports, see Recommended Microphones.
- Keep background noise to a minimum.
Background noise can affect accuracy, as it may be mistaken for a word. Excessive background noise, like loud music or television, a phone ringing, text message notification noises, or talking in the background may affect dictation results. For best results, try not to dictate with competing sounds in the background.
- Speak at a normal, consistent rate.
Speak in a smooth, non-rushed manner, avoiding rushing over smaller, connecting grammar. Also avoid speaking too slowly and haltingly, such as while waiting to see your transcript return before you continue speaking. Articulate all words distinctly and cleanly, including punctuation. Don't blend or slur word boundaries. Speaking in complete sentences using expected speech patterns will help to produce the best results.
- Stop the recording between long pauses.
Think about what you want to say before you begin dictating. This can smooth out the overall speech pattern, and reduces “disfluencies” (the “thoughtful spoken pause” commonly heard as an “um” or “ah”). Also, remember that nVoq dictation servers can only transcribe what you say, not what you meant to say.
Tips for Accented Speakers
In addition to all of the tips above for all speakers, accented speakers may present increased challenges for speech recognition. Problematic accented speech includes languages in which certain consonants have very different or non-existent presentations, vowels are less distinct, or extra syllable-like vocalizations are inserted. While the human ear may have no problem interpreting accents, a significant portion of comprehension is due to deep contextual knowledge that automatically corrects and deletes audio input that is not meaningful. Speech Recognition must account for all audio input, and has limitations on deep context. That being said, many ESL (English as a Second Language) speakers do quite well with dictation and have learned to speak specifically for voice recognition accuracy.
- Practice speaking for an accurate return while keeping in mind the few distinct challenges related to the particular accent. This may sound awkward or unnatural at first, and will probably not be how a speaker would communicate with another person.
- Articulate all words distinctly and cleanly, and try not to drop connecting grammar such as articles, prepositions, and past tense verb endings.
- Speak in a smooth, non-rushed manner, but avoid speaking too slowly and robotic. This is true for all speakers, but is often heard in accented dictation recordings.
Not all speakers are able to alter an accent. Some additional support for specific account level vocabulary, sentence modeling, or substitutions may help, but please be aware this may be an on-going and time-consuming process. This is not a commentary on a speaker’s ability to communicate effectively or speak fluently, but simply an acknowledgement of the limitations speech recognition.
Understanding and communicating these challenges in a non-confrontational manner can save time and allow clients to either patiently work with speech recognition, or make an informed decision to look for other solutions such as working with a scribe or transcriptionist.