Troubleshooting

Check that your microphone is properly connected.

Ensure Speech Recognition Cloud has permission to access your microphone in Windows Privacy settings (Settings > Privacy > Microphone).

Try selecting a different microphone in SRC Settings if you have multiple audio devices.

Test your microphone in another application to verify it is working.

See our tips for better accuracy page for detailed guidance.

Add custom vocabulary for names and technical terms in Settings.

Try speaking more clearly and at a steady pace.

Reduce background noise or switch to a headset microphone.

Set up text replacements for words that are consistently wrong.

Add the word to your custom vocabulary in SRC Settings. This is especially important for names, technical terms, and industry-specific jargon.

If the word is still wrong after adding vocabulary, set up a text replacement rule to force the correct spelling. Add multiple variations to catch different ways the word might be mistranscribed.

During periods of silence or when the microphone does not pick up clear speech, the transcription model may produce unexpected phrases. This can happen with any cloud speech recognition system.

To reduce this: use a quality microphone positioned close to your mouth, check your microphone volume is not too low, reduce background noise (TV, music, other voices), and avoid long pauses while recording.

The Ultra Accuracy mode (available on Professional and Medical plans) significantly reduces this behaviour.

The transcription model may occasionally output text in a different language, especially with short audio segments.

To reduce this: select your specific language in SRC instead of using auto-detect, add vocabulary in your preferred language, and record longer segments as the model is more accurate with more audio context.

Ensure you are speaking in complete sentences. SRC uses full-sentence context for punctuation, so very short phrases may not receive correct punctuation.

You can also use spoken punctuation commands to insert punctuation manually when needed.

Some sound cards have a slight delay when activating the microphone. If you notice the beginning of your speech being cut off, try switching to a different microphone or sound card, or waiting half a second after pressing the hotkey before speaking.

Some sound cards add a noticeable delay before recording starts. This is often caused by built-in audio processing such as noise reduction, echo cancellation, or audio enhancements.

Check your sound card or audio driver settings and disable any audio processing or enhancements. On some systems, look in the Realtek, Intel, or manufacturer-specific audio control panel.

Speech Recognition Cloud works well even with background noise, so disabling noise reduction is usually fine. If the delay persists, try a USB microphone (they often have less processing overhead).

Check your internet connection. Speech Recognition Cloud requires a stable connection as processing happens in the cloud.

Close other bandwidth-heavy applications that may be competing for network resources.

Check that no other application is using the same hotkey.

Try setting a different hotkey in Settings.

Fully restart Speech Recognition Cloud: right-click the tray icon and select Exit, then start the app again.

If that does not help, restart your computer.

Speech Recognition Cloud inserts text wherever your cursor currently has focus. Make sure the correct application window is focused and your cursor is in the right text field before you start dictating.

If this happens, you can press Ctrl+Alt+V to paste the last transcription into your target application.

If Speech Recognition Cloud types out text that was previously in your clipboard instead of what you said, try restarting your computer. The Windows clipboard can slow down over time, and a restart clears this up.

Make sure you are running Windows 10 or Windows 11.

Right-click the installer and select "Run as administrator".

If your antivirus blocks it, add an exception for the installer. Some antivirus software may flag Speech Recognition Cloud because it uses keyboard simulation to type text. This is a false positive and the software is safe to use.

Check your system tray (bottom-right corner of the taskbar). SRC runs as a system tray application. Click the small up arrow to see hidden tray icons.

Some antivirus software may flag Speech Recognition Cloud because it uses keyboard simulation to type text. This is a false positive. The software is safe to use. You may need to add it to your antivirus whitelist or create an exception.

Sign out of SRC and sign back in.

If the issue persists, check that your subscription is active in your account settings (Settings > General > Manage Subscription).

If you still have problems, contact us and we will sort it out.

Open Speech Recognition Cloud, go to Settings > General, and click "Manage Subscription". Under Invoice History you can view and download all your invoices as PDF.