In this guide
The quick answer
First identify the tool you’re using: macOS Dictation, Echo, or a browser’s voice typing. Then test a short sentence in a simple text field. Check the shortcut if nothing starts, the microphone if nothing is heard, and permissions or focus if text cannot be inserted.
Find the stage that failed
Use a short test sentence in a new TextEdit document. A plain local document removes browser permissions, document access, and rich text controls from the first test. Keep other dictation tools stopped while you diagnose.
| Symptom | Check first |
|---|---|
| No listening indicator | The app is running, the shortcut is correct, and required permissions are enabled. |
| Listening, but no useful transcript | The selected microphone, input level, language, and model status. |
| Correct words, long wait | The cleanup mode, provider connection, and local model state. |
| Transcript appears, text does not | Cursor focus, edit access, and Echo’s Accessibility permission. |
For the Dictation built into macOS
Open System Settings → Keyboard → Dictation. Confirm it is enabled, check the configured shortcut, select the intended language, and check the microphone source. In Sound → Input, make sure the input responds when you speak.
Apple recommends checking microphone placement, input volume, background noise, and whether your configuration needs an internet connection. Its Dictation troubleshooting page explains those checks. Use the processing notice below Dictation in Keyboard settings for your Mac’s network requirements.
For Echo: the shortcut does nothing
First look for Echo in the menu bar. Open Settings → General and check Hold to talk. Hold that key briefly, speak after the listening indicator appears, then release. A quick tap is not the same as holding to dictate.
If fn opens the emoji picker, choose System Settings → Keyboard → Press 🌐 key to → Do Nothing, or select another key such as Right Option in Echo. Avoid giving two dictation tools the same shortcut.
Open Settings → Permissions in Echo. Check Microphone and Accessibility. Input Monitoring is only needed if the shortcut does not work with Accessibility alone; Echo’s permission view helps you identify what is missing. Follow any macOS prompt to reopen the app after a permission change.
Echo listens, but the words are wrong or missing
In Settings → General, check the selected microphone. An attached headset or display can change the input you expected to use. Watch the input level while speaking, and try the Mac’s built-in microphone if you have one.
In Settings → Speech, check that the speech model is ready. A model that is still downloading cannot complete the same job as a prepared one. Confirm the dictation language; if Automatic struggles with a short phrase, choose a language explicitly and try a complete sentence. The language guide explains that setup.
The transcript is right, but the result is late or absent
Temporarily choose Settings → Polish → Off. If the delay disappears, recognition is working and the cleanup stage needs attention. A cloud provider needs a working connection and account; a local AI model needs its server running. Echo falls back when cleanup fails, but a working offline mode avoids waiting on an unavailable provider.
If the words appear in Echo but not in the destination, click in an editable text field and try again. Check Accessibility access. Password fields are a special case: Echo copies text instead of automatically pasting into them.
When History is enabled, retrieve the text there rather than dictating the whole passage again. If the simple TextEdit test works but one web editor fails, narrow your checks to that editor’s focus and permissions. For Google Docs, also confirm you can edit the document.
If you still need help, contact info@bitprods.com with your Echo version, macOS version, selected language, and the stage that fails. A short description of the test is usually more useful than a private document or a long transcript.