How to fix common HoldSay problems
Trace common failures to the setting at fault.
Start with the exact symptom instead of changing every setting at once.
The checks below separate shortcut, cursor, audio, model and source file problems so you can test one cause at a time and confirm whether the issue is fixed before moving on.
Start from the symptom — each one points at the permission, device or model to check.
The shortcut does nothing
You are ready to dictate into a note, message, or document, but holding the shortcut does not start dictation. First, open HoldSay and look at the floating card. It needs to be ready before it can handle held key dictation. The default shortcut is Right Option on Mac and Right Ctrl on Windows, but your configured shortcut may be different.
- Open Settings in HoldSay and confirm the configured dictation shortcut.
- Click inside a plain editable text field and test the shortcut there. This separates a HoldSay shortcut issue from a limitation in the app where you first tried to dictate.
- If the shortcut still does nothing, check whether another open app is using the same shortcut.
The held dictation shortcut is visible and changeable in Settings.
On Mac, also recheck Accessibility and keyboard permissions. These permissions affect whether HoldSay can respond to your keyboard input and place dictated text in another app. If you change a permission, quit and reopen HoldSay before testing again. Return to the same plain text field for the retest, so only one condition has changed.
The card listens but no text appears
The floating card can show that it is listening while the destination app receives no text. In this situation, the most important condition is the insertion point. HoldSay needs an active editable destination when dictation begins. Looking at a document or conversation is not enough if the text cursor is not active.
- Click the exact field where you want the words to appear.
- Confirm that a blinking text cursor is visible.
- Start one short dictation, then wait for its processing and insertion to complete before beginning another one.
A finished dictation is placed where the cursor is already active.
If the cursor was visible but text still does not arrive, confirm the microphone selection. A headset reconnecting can change the selected input, even if the headset was working earlier. Check the dictation language as well. Select the language you are actually speaking, then repeat the same short test phrase.
Do not start a second dictation while the first result is still being processed. Multiple attempts can make it unclear which result belongs to which spoken phrase. For a controlled test, use a short sentence in a plain text field, keep the same microphone connected, and wait until the current result has either appeared or finished processing.
A recording is missing one side
A call recording can be incomplete in two different ways. Your own voice comes from the microphone. Other people in a computer call come from computer audio and depend on the output device used by the meeting app. Treat these as separate checks, because fixing one does not confirm the other.
If your voice is missing, recheck the microphone selection and the relevant operating system permission. This is especially important after connecting or reconnecting Bluetooth headphones, wired headphones, speakers, or another microphone. The device you expect to use may no longer be the selected input.
The source label confirms that HoldSay is recording your microphone and the sound from the computer.
If the other people on a call are missing, recheck computer audio capture and the output device selected in the meeting app. A meeting app can send call audio to a different output device after a Bluetooth, headphone, or speaker change. Make a short call recording after any such change rather than waiting until an important meeting begins.
- Start a short test call with speech from both sides.
- Use the floating card and REC to make a brief recording.
- Review the recording and confirm that your voice and the voices from the call are both present.
Repeat the test whenever you change a headphone, Bluetooth, speaker, microphone, or meeting output device. The goal is not only to see that recording started, but to verify both audio sources before relying on the setup.
A model or file transcription will not finish
When a model download or file transcription does not finish, first identify whether the problem is the model, the source media, the drive containing it, or available disk space. Changing several of these at once makes the result harder to interpret.
For an initial model download, keep the computer online until the download finishes. If the source is a media file, open that file in a normal player. This confirms that the media itself opens before you retry transcription in HoldSay.
A failed row keeps the source visible so you can fix the cause and retry.
- Confirm that the computer remains online if the initial model download has not completed.
- Open the source media in a normal player.
- If the file is on an external or network drive, reconnect that drive and confirm the source is available.
- Restore enough free disk space.
- After fixing the relevant cause, choose Retry in the file queue.
Use the same source file for the retry when possible. If it opens normally, the drive is connected, sufficient disk space is available, and the retry still does not finish, record the exact error message for support. Do not assume that a file is unavailable just because it was previously accessible. External and network sources can become disconnected after the file was added to the queue.
Send support enough information to reproduce it
If the controlled checks do not resolve the issue, send support a report that shows the exact condition rather than a general description. Before writing, open Settings, choose About, find Health Check, and choose Open.... The report gathers app state that is useful for diagnosis without requiring you to describe every internal component yourself.
- Open Health Check and save or collect the report it provides.
- Fix the first failed item if Health Check identifies one, then repeat the same short controlled test.
- Email support@holdsay.com with your platform, HoldSay version, Health Check report, exact action, expected result, actual result, and the full error message.
Include the operating system, HoldSay version and exact failed step in a support report.
Describe the test precisely. For example, state whether you held the configured shortcut in a plain text field, whether a blinking cursor was visible, which audio side was missing, or whether the source file opened in a normal player. Include what appeared on screen instead of only saying that it failed.
A short screen recording can help when it does not expose private audio or license information. Do not send an important confidential recording unless support specifically asks for a safe diagnostic sample. A short test file that produces the same issue is more useful than a confidential file because it lets support investigate the behavior without unnecessary private material.
Common questions
Why is my HoldSay keyboard shortcut not working?
Confirm that HoldSay is open, check the shortcut in Settings, and test it in a plain text field. On Mac, also recheck Accessibility and keyboard permissions, then quit and reopen HoldSay after changing a permission.
Why does HoldSay listen but not insert text?
Click an editable text field so a blinking cursor is visible, then confirm the selected microphone and dictation language. Wait for the current dictation to finish processing before starting another one.
What should I include in a HoldSay support request?
Include your platform, HoldSay version, Health Check report, exact steps, expected result, actual result, and the full error message. Use a non confidential test file or short screen recording when possible.
If a label on this page does not match the app, email support@holdsay.com. Include Mac or Windows and the HoldSay version shown in Settings.