Skip to content
Petal

FixesMac6 min read

Superwhisper empty transcription: why the result is blank

Superwhisper empty transcription? Check the microphone and waveform, play the recording in History, confirm the model download, then retry the audio.

Published by Optimal Apps, the maker of Petal.

On this page

A Superwhisper empty transcription almost always comes from the audio input, not the model. When History shows "No voice found in recording", Superwhisper says the recording had no detectable speech. Check the selected microphone, watch the waveform while you speak, and play the recording back in History. If you hear your voice clearly, the problem is the model or processing: confirm that the model finished its download, restart Superwhisper, and retry the dictation from History.

First, make sure you have an empty result and not a paste problem. The two look the same from the text field.

Quick-fix checklist

What you see Likely cause Fix
Recent recordings shows the text, field is empty Paste failed, not transcription Follow the paste guide
"No voice found in recording" No speech reached the app Check the microphone and permission
Waveform stays flat while you speak Wrong or muted input Select another mic or Use system default
Playback in History is silent Input was wrong or muted Fix the input, then record again
Playback is clear, result still empty Model or processing problem Check the model download, restart, retry
Stuck on processing Cloud model offline, or model still downloads Check the connection; try a local model
No History entry at all Recording did not start or was discarded Check the shortcut, mic permission, and mode

Step 1: Separate a Superwhisper empty transcription from a paste problem

Open Home in Superwhisper and look at Recent recordings.

  • The dictation is listed and has text. Transcription worked. The text did not reach your app. Read Superwhisper not pasting text instead.
  • The dictation is listed but is empty, or shows "No voice found in recording". This guide is for you. Go to Step 2.
  • There is no History entry at all. The recording did not save. Go to Step 6.

Step 2: Check the selected microphone

Superwhisper records from the microphone that it has selected. This is not always the device you expect.

  1. Click the microphone button at the top right of the Home, Configuration, or Sound page in Superwhisper.
  2. If a disconnected headset or a sleeping Bluetooth device is selected, switch to a working device, or choose Use system default.
  3. Open System Settings > Sound > Input, speak, and confirm that the input level moves for the same device.
  4. If the input level is low, drag the Input volume slider to the right.

Why it works: if Superwhisper records from a device that sends no audio, the model gets silence, and silence gives an empty result.

Bluetooth headsets cause many of these cases. When a headset switches to its microphone mode, the audio quality drops, and some devices send almost nothing. Try the Mac's built-in microphone to rule this out. For more, read Bluetooth microphone dictation on Mac.

Step 3: Confirm the microphone permission

  1. Open System Settings > Privacy & Security > Microphone.
  2. Make sure Superwhisper is on. If it is on, turn it off and on again.
  3. Quit and reopen Superwhisper.

Without this permission, macOS gives the app silence instead of audio. Read Mac microphone permission if Superwhisper is not in the list.

Step 4: Watch the waveform, then play back the recording

  1. Start a recording and speak a full sentence.
  2. Watch the waveform. If it stays flat, audio is not reaching the app. Go back to Step 2.
  3. Open the dictation in History and play it.

The playback tells you where the fault is:

  • Silent playback. The input was wrong or muted. Fix the input, then record again.
  • You hear yourself clearly, but the result is still empty or "No voice found". The audio is fine. Report the recording from History so Superwhisper can see it, and go on to Step 5.

One exception: modes that record system audio can show a flat waveform. That is expected for those modes.

Also turn on Automatically increase microphone volume in Settings > Sound. It raises the input volume when a recording starts. It works only when Superwhisper uses the system default device.

Step 5: Check the model and the processing step

If the audio is good, the model step is the next place to look.

  1. Check the connection. Cloud models need the internet. If you are offline, switch the mode to a local model to test.
  2. Check the model download. Open Settings > Models library and make sure the model finished its download. A model that is still on its way cannot transcribe.
  3. Restart Superwhisper, then retry the dictation from History. Superwhisper keeps the audio, so you do not need to speak again.
  4. Turn on Error logging in Settings > Configuration, reproduce the problem once, and report the recording from History.

Intel Mac speed limits

Superwhisper's FAQ answers "Will it work on my Intel Mac?" with: Intel Macs work best with cloud models, and offline models only run really well on Apple silicon Macs. Some local models, such as Parakeet, are not available on Intel. On an Intel Mac, a large local Whisper model can take a long time, which can look like a stuck or empty result. Use a smaller model, or a cloud model, on Intel.

Step 6: Fix a Superwhisper recording with no History entry

If the dictation does not appear in History at all:

  1. Check that your shortcut starts a recording. You should see the recording window.
  2. Check that you did not cancel the recording with Escape.
  3. Confirm the microphone permission (Step 3).
  4. Open the mode you use and make sure it has a model you can access. A mode with no usable model can fail before it saves a result.
  5. Update Superwhisper from the menu bar icon with Check for Updates....

A user report to keep in mind

In a published review on tl;dv, a reviewer said a 25-minute recording came back as "No voice found". The reviewer then dug through the app's files and got the content back. Treat this as a symptom, not a verified fix: it shows that an empty result does not always mean the audio is gone. Before you delete anything, play the recording in History and retry it.

Why empty results happen in any dictation app

Every dictation app runs the same three steps: capture the audio, run the speech model, and insert the text. An empty result means step one or step two failed:

  • Capture: wrong device, muted input, no permission, Bluetooth mode switch.
  • Model: missing download, no connection for a cloud model, a model too slow for the Mac, or a speech detector that decided the clip had no speech.

When you know which step failed, the fix is usually short.

How Petal handles a bad transcription

If you want to compare, Petal for Mac is a free, open-source dictation app from Optimal Apps, the publisher of this site. It runs on-device speech models and keeps each recording, so a bad result is easy to redo.

  • Pick the microphone in one place. Settings > Recording > Microphone lists your input devices, with Default for the system device.
  • The audio is kept. By default, Petal saves each recording with its transcript in ~/Documents/petal/history/.
  • Transcribe Again. On a History card, Transcribe Again runs the saved audio through the model again. Switch to a different model in Settings > Models first if you want a second opinion.
  • On-device models with clear sizes. Petal lists each model with its size, such as Parakeet 110M (the recommended light English model) or Whisper Large V3 Turbo (about 1.1 GB, 99 languages). Petal keeps the RAM need of its default model under 8 GB.
  • No cloud step for speech. Speech recognition runs on your Mac, so a dropped connection cannot empty a result.

Petal needs macOS 26.2 or later. For a side-by-side comparison, read Petal vs Superwhisper.

Summary

  1. Check Recent recordings to confirm that the result is empty and the paste is not the problem.
  2. Select a working microphone and confirm the input level in Sound settings.
  3. Turn the microphone permission off and on.
  4. Watch the waveform, then play the recording in History.
  5. If the audio is clear, check the model download and connection, restart, and retry from History.
  6. On an Intel Mac, use a smaller or cloud model.

Questions

What does No voice found mean in Superwhisper?

Superwhisper says it means the recording had no detectable speech. The cause is usually the audio input, such as the wrong microphone or a muted device, not the model.

How do I know if the problem is the transcription or the paste?

Open Home and look at Recent recordings. If the dictation is listed with text, transcription worked and the paste failed. If it is missing or empty, the problem is the microphone, the model, or processing.

Can I get the text back from an empty Superwhisper recording?

Superwhisper saves the audio. Play it in History. If you hear your voice, fix the model or connection, restart Superwhisper, and retry the dictation from History.

Why is Superwhisper slow or empty on an Intel Mac?

Superwhisper's FAQ says Intel Macs work best with cloud models, and offline models run really well only on Apple silicon. A slow local model on Intel can look like a stuck or empty result.

FixesMac

Mac dictation wrong language: how to fix it

Mac dictation wrong language? Add your language in Keyboard > Dictation, switch it while you dictate, check the input source, and finish the download.