Why is my Speech to Text accuracy low?

Overview

This article explains common causes of low speech to text accuracy and how to address them.

Applies to

All Users

Poor audio quality

The most common cause of low accuracy is poor audio quality. Use a dedicated USB headset or professional speech microphone rather than a built-in microphone. Ensure the microphone is positioned correctly and that you are recording in a quiet environment.

Background noise

Background noise, such as office chatter, HVAC systems, music, or other ambient sounds, significantly reduces accuracy. Dictate in a quiet space where possible.

Speaking too quickly or unclearly

Speaking too fast, mumbling, or failing to pronounce words clearly reduces recognition accuracy. Speak at a natural, measured pace with clear diction.

Missing specialist vocabulary

If the AI engine consistently misrecognises specific terms particularly medical, legal, or industry specific words add them as Speech to Text Replacements.

Incorrect language selected

Ensure the correct language is selected in your Dictalogic settings. If English (US) is selected and you are dictating in English (UK), accuracy may be affected for certain vocabulary.

New user adjustment period

The AI engine adapts to your voice over time. Accuracy typically improves with continued use. Be consistent in your microphone setup and dictation environment.

Contact support

If you have addressed the above factors and accuracy remains poor, contact Dictalogic support for further investigation.