Transcribble Help

Getting started

1. Drop an audio file onto the Transcribble window, or click Choose File… to pick one.
2. The first time you transcribe in a language, macOS downloads a small speech model. This happens once; after that, everything works offline.
3. The transcript streams in live while the file is processed, with a progress bar.
4. When it finishes, the full text appears. Edit it in place if you like, then click Copy Transcript to put it on the clipboard.
5. Click New to start over with another file.

Supported audio formats

Transcribble reads any format macOS can decode, including:

• MP3
• M4A / AAC (including Voice Memos recordings)
• WAV
• AIFF
• CAF
• FLAC

Ogg and Opus files are not supported by the system decoder. Convert them to one of the formats above first (for example with any free audio converter).

Languages

Transcribble automatically uses the speech model matching your Mac's system language. Apple's engine supports major world languages, with more added in macOS updates. If your system language isn't supported, Transcribble falls back to the closest available language, then to English.

Privacy

All transcription happens on your Mac. Your audio and transcripts are never uploaded anywhere. The only network activity is Apple's one-time speech model download, which comes from Apple, not from us. See the full privacy policy.

Troubleshooting

"Couldn't transcribe that file"

• Check the file is an audio format from the list above. Video files and Ogg/Opus audio are not supported.
• Very short clips (under a second) or files with no speech may produce an empty transcript.
• If the file is DRM-protected (for example, purchased audiobooks), macOS cannot decode it.

The model download seems stuck

The one-time language model download needs an internet connection and can take a few minutes on slow connections. Check your connection and try again — the download resumes automatically.

The transcript has mistakes

Accuracy depends on recording quality. Clear speech close to the microphone transcribes very well; distant, noisy, or heavily overlapping speakers less so. The transcript is editable in the app, so you can fix anything before copying.

The app won't open

Transcribble requires macOS 26 (Tahoe) or later — the Apple speech engine it uses doesn't exist on earlier versions of macOS.

Frequently asked questions

Does it work offline?

Yes. After the one-time model download, no internet connection is needed.

Is there a length limit?

No hard limit. Long recordings simply take longer — typically far less time than the recording's duration.

Does it identify different speakers?

Not currently. The transcript is continuous text without speaker labels.

Can it record audio or transcribe live?

No — Transcribble deliberately does one thing: turn existing audio files into text. For live dictation, use the built-in macOS dictation feature.

Contact

Questions, bug reports, or feature requests: email rik.watson@gmail.com.