Yes! Under the advanced options, you can set the minimum and maximum lengths of each caption or title.

Version 2.0 includes nearly 100 languages via the Whisper engine. Click here for details: Multi-Language Support

No, Captionator runs entirely on your Mac. A one time purchase will work forever!

There is a bug with the large language model where it won’t download properly from the server. The model URL changed on the upstream server. There is a fix in the next version, but you can also use the Medium model, which has a very similar accurac.

On some Macs, if you use Whisper with CoreML, it doesn’t error out, nor does it ever finish processing. I am working for a more permanent fix, but there is a quick solution by downloading a different model and not downloading the CoreML addon for that model. Once done, it should process normally.

See this video for a tutorial on how to make this change.

Captionator includes two transcription engines called Whisper and Siri. The Siri engine only supports some languages on-device. Languages that aren’t supported on-device require cloud transcription that imposes a 60 second limit.

It is best to use the Whisper engine, found under the preferences. All of the Whisper languages are supported on-device, and thus have no limit.

Captionator adds Whisper to dramatically increase accuracy. You can enable Whisper in the preferences. You can also install more accurate models that run slower. By default, the “tiny” model is installed, but you can install the Small, Medium, or Large models to make the transcriptions much more accurate. See the support video for details:

Using Whisper with Captionator

Make sure to disable all music of SFX layers before you press the Export button. Speech to Text can get easily confused if there is music playing or sound effects break up the words

Yes! Once the titles are on your timeline, move one title to the desired location and Copy (Command-C), then select the rest of the captions and select Paste Attributes… (Command-Shift-V), then select the transforms you want. Note: This MUST be done in the transforms panel, as Final Cut doesn’t currently copy the text location from the Text panel.

Another technique is to highlight all the titles you want to move, then click the visual transform button below the preview pane (the square with 4 corners). Move the title to the desired position, and the remaining titles will follow.

Use the roles at the bottom left. Click “Index”, then “Roles”. Expand “Titles”, and enable/disable “Generated Titles”

Click on the Help, then “Captionator Help”, then follow the directions. Use the “Install Export Destination” on the second page for an express install.

No! We feel strongly that collecting your data is totally unnecessary for this type of application. As such, we have worked hard to ensure no data is collected. However, as the app is available on certain app stores, they may be collecting information from you. Please check their privacy policies for more information. Ours is here: Privacy Policy

These fonts are licensed, so be sure to check licensing requirements, as you would with other fonts.

SF Pro Display

Trash Hand

Komika Axis

Final Cut has a bug that sometimes prevents it from importing valid, modern FCPXML files. We believe it’s related to a faulty time machine backup, but have been unable to track down the root cause.

The fix, however, is pretty simple. Open Captionator’s preferences, then change the “DTD Ver” to something older. Anything from 1.8 forward should work for most people, but sometimes new features in newer Motion templates aren’t supported.

This is usually caused by using Apple’s cloud transcription service. Under Preferences, there is a cloud setting; you can change this to Local Only, or Allow Cloud. This will make Captionator use your local transcription engine instead of the cloud, and should clear this error.

Alternatively, switch to the Whisper engine, which eliminates this problem entirely.

This is how Whisper tells us that it can’t understand the audio. The built in tiny model works in many cases, but if the audio isn’t nearly perfect, it can get confused. Fortunately, adding a larger model, such as the Small or Base model, should solve this.

See here for details