Advanced Settings

Reference for HyperWhisper’s advanced recording and storage settings — max duration, audio sample rate, audio file retention, and history limits.

Advanced settings control low-level recording behavior and data-retention limits. They are stored in HyperWhisper’s preferences and are included in backup files so they travel with the rest of your configuration.

These settings are not exposed in a dedicated Advanced panel in the current Settings UI. You can view and change them through the backup import/export workflow described below, or by editing a .hwbackup.json file before importing it.

---

Settings reference

Max recording duration

Controls the maximum length of a single recording session.

Detail	Value
Default	300 seconds (5 minutes)
Platform	macOS

When a recording reaches this limit, it stops automatically and processing begins on the audio captured so far. This prevents runaway recordings from accumulating large audio files unexpectedly.

---

Audio sample rate

The sample rate (in Hz) at which HyperWhisper captures microphone audio for standard (non-streaming) recordings.

Detail	Value
Default	16 000 Hz
Valid range	8 000 – 48 000 Hz
Platform	macOS

If a value outside the valid range is stored (for example, from a corrupted preference), it resets automatically to 16 000 Hz at launch. The default of 16 000 Hz is the standard input rate for Whisper-family models and matches what most cloud STT providers expect.

Streaming transcription uses a fixed sample rate set by the streaming provider, not this setting. See Streaming Settings for details.

---

Keep audio files

Controls whether the raw audio recording is kept on disk after a transcription completes.

Detail	Value
Default	Off (audio is discarded after transcription)
Platform	macOS

When off, audio files are deleted once transcription succeeds, which saves disk space. When on, the audio file is retained alongside the transcript in your recordings folder and can be played back from the History view.

Turning this on also enables Retry on failed transcriptions, because the audio file needs to be present on disk to retry. See Managing Your Recording History for playback details.

---

History retention days

The number of days to use as the default threshold for history-based operations.

Detail	Value
Default	30 days
Platform	macOS

If a negative value is stored, it resets to 30 at launch. This value is separate from the auto-delete feature — it represents a preference value that can be carried across devices via backup. See Managing Your Recording History for automatic cleanup configuration.

---

Show experimental models

When enabled, experimental and preview model variants appear in the model selector alongside stable models.

Detail	Value
Default	Off
Platform	macOS

Experimental models may produce better results on specific content types but can be less stable or subject to removal in future updates.

---

Changing advanced settings via backup

Because these settings do not appear in a dedicated Settings panel, the most reliable way to change them is through the backup workflow.

1. Export your current settings

   Go to Settings → Backup, make sure Settings is included, and click Export. Save the .hwbackup.json file somewhere you can find it.

2. Edit the file

   Open the .hwbackup.json file in a text editor. Find the "advanced" object near the bottom of the "settings" block:

   "advanced": {
     "maxRecordingDuration": 300,
     "audioSampleRate": 16000,
     "keepAudioFiles": false,
     "historyRetentionDays": 30
   }
   

   Change the values you want to adjust. Make sure the file remains valid JSON before saving.

3. Import the edited file

   Go to Settings → Backup and click Import. Select the file you just edited. Review the import summary and confirm. Your new values take effect immediately.

The showExperimentalModels setting is stored under the "aiModel" object in the backup file, not under "advanced".

---

Windows

The settings described on this page (maxRecordingDuration, audioSampleRate, keepAudioFiles, historyRetentionDays) are macOS-only. Windows does not implement them: they are not surfaced in the Windows UI, not written when Windows exports a backup, and not applied when Windows imports a backup. If a macOS backup file containing these fields is imported on Windows, the fields are silently ignored.