When voice-to-text is not working, first check whether the tool is receiving your voice, producing a transcript, or placing that text where you want it. Those are different jobs. A microphone problem needs a different fix from a language mismatch or a text field that is not ready for input.
Start with one short, non-private sentence and one change at a time. If you already have a useful recording, preserve it before retrying. Repeating an important thought from memory is a poor substitute for recovering the original audio.
Match the symptom to the first failing step
Use this table to choose a test before changing settings.
| What you see | What to check first | A useful comparison |
|---|---|---|
| No microphone button | Whether this input method offers voice typing and is enabled | Open a normal text field in an app supported by that input method |
| A microphone button, but no sign of captured sound | Selected microphone, connection, mute state, and permissions | Record a short sample in a basic recorder and replay it |
| Audio is audible, but no transcript appears | Language, recognition status, connectivity, and the actual error | Try one short sentence in the same tool |
| Words appear in the voice tool, but not the destination | The active text field and the tool's insertion or copy workflow | Try an ordinary note, then copy and paste the existing text |
| Text appears, but important words are wrong | Spoken language, audio clarity, and the original audio | Check a name and a number against the recording |
| An existing audio file is rejected | File format, size, and the service's file-upload path | Confirm the file plays locally; keep the original |
| A transcript exists, but the polished output is missing | The separate cleanup or formatting step | Inspect the transcript before repeating capture |
A comparison narrows the problem; it does not prove a cause by itself. A basic recorder and a dictation tool may use different permissions or microphones. Record which test worked and which failed so the next step remains specific.
Run a small capture-to-text check
Here is an illustrative test sentence:
My next task is to review note seven on Friday.
It gives you a normal phrase, a number, and a day to inspect. It is a sample for your own check, not a claim about any app's accuracy.
- Choose the input method you intend to use. A keyboard microphone, an operating-system dictation shortcut, a browser recorder, and a file-upload service are separate entry points.
- Speak the sample once. Note whether the tool shows activity, returns text, or displays an error.
- If you recorded audio, replay it. If you have text, check the number and the day.
- Change one relevant setting and repeat the same sample.
- Stop when you have enough evidence to name the first failing step. Do not keep resetting unrelated settings after a narrower test has identified the next thing to inspect.
For an existing recording, use a service that accepts audio files. Live dictation is not automatically a file-transcription tool. The guide to converting a voice memo to text explains that separate starting point.
Check the input method on your platform
The following checks refer to each provider's documented input method. They do not establish that every third-party app has the same features, settings, or processing behavior.
iPhone keyboard Dictation
If the keyboard's dictation control is missing, check whether Dictation is enabled under General and Keyboard in the Settings app. Then place the insertion point inside a text field before starting it. Apple also ties dictation language to the keyboard language, so check the keyboard you are actually using.
Use Apple's iPhone Dictation guide for your software version. A voice message button inside an app may record audio instead of inserting text; identify that button's job before treating the result as a dictation failure.
Android with Gboard
For Gboard, check its microphone permission and whether Voice typing is enabled in the keyboard's settings. If you use advanced voice typing, Google's troubleshooting guidance also calls out the relationship between the Gboard language and the device language. These advanced options are not identical on every Android device.
Follow Google's current Gboard troubleshooting instructions for the feature and device you use. Begin with the permission, enabled feature, and language checks. A problem with Gboard does not establish that a separate recording app is broken.
Windows voice typing
Check the selected input device under System and Sound, and microphone access under Privacy & security. If Windows reports an unsupported voice-typing language, check its supported-language guidance. For incorrect words, Microsoft also recommends checking the active language and the recording environment.
Use Microsoft's Windows voice-typing troubleshooting page for the current steps. If no text appears, confirm that the microphone input is usable before interpreting the symptom as poor recognition. Windows voice typing, voice access, and an app's own Dictate button are different features; troubleshoot the one you actually started.
macOS Dictation
Open Keyboard settings and inspect Dictation, its selected language, shortcut, and microphone source. The microphone chosen for Dictation deserves its own check even if another app can record sound.
Apple's Mac Dictation guide also explains how to inspect whether your configuration needs an internet connection. Do not assume every voice workflow is either fully offline or always online. Match the check to the documented input method and your current configuration.
A voice tool inside Chrome
A website using your microphone needs the relevant site permission as well as access allowed by the operating system. Check the permission for the exact site you are using and, where available, its selected microphone. Access granted to a different website does not settle this check.
Use Google's Chrome microphone-permission guide. If the page displays an error, keep its wording and the time of the attempt. A permission check alone does not prove that the website's recognition or processing service is available.
If text exists but does not reach the destination
Look for the existing transcript before recording again. A successful recognition result and successful insertion into another application are separate outcomes.
Try an ordinary, editable note field. If that works but your intended destination does not, record that difference. It gives support a narrower reproduction than “voice typing never works.” Check whether the tool expects you to insert, copy, export, or save its result rather than assuming that recording sends text everywhere automatically.
For a temporary workaround, copy text that already exists into the destination and inspect what arrived. This preserves work while you investigate the original insertion path; it does not establish that the original path is fixed.
Preserve the source when processing fails
If a recording matters, confirm where the actual audio is stored before starting a destructive recovery step. Avoid deleting app data or reinstalling software as an initial response to a missing transcript. Those actions can have consequences unrelated to the failure you are trying to locate.
A transcript can also be available while a later summary, cleanup, or formatting step is still pending or has failed. Inspect each result separately. Fluent rewritten text is not evidence that names, numbers, or decisions survived correctly; check them against the source when it is available.
If you are using Vowise
Vowise's recording and transcription troubleshooting guide distinguishes capture, processing, and output failures. It explains that a stopped Desktop recording with captured audio is retained in local Records before later processing finishes. Check that retained item before making a replacement recording. This does not mean every feature or platform works fully offline.
For a new capture, follow the first-recording guide to check the selected input, language, and account access. After you have usable text, decide whether you want a note, an action, or a personal reflection. The Journal & Review workflow is a next step for reflective entries; it is not a repair for another product's microphone or keyboard.
Keep a support note that helps someone reproduce the problem
Use this outline with non-private details:
- Input method: operating-system Dictation, keyboard voice typing, browser recording, or file upload.
- Device and software: the relevant operating system, app, or browser version.
- Expected result: audio saved, transcript produced, or text inserted into a named field.
- Observed result: the first step that failed and the exact error, if one appeared.
- Time of the attempt: include your time zone.
- Comparison: what happened with the same short sample after one change.
- Source preserved: whether the audio or existing text can still be reopened.
Leave passwords, keys, private recordings, and sensitive transcripts out of screenshots and logs. A useful support note explains the failure without exposing the material you were trying to capture.
You have a working result when the intended input produces usable text, that text reaches the place you need it, and important details survive review. If one step still fails, keep the successful parts and investigate that step rather than beginning the entire workflow again.