Transcribe a file

Transcribing a file you already have is the quickest way to start. Everything runs on your Mac, and you need at least one model (Whisper Tiny is free, see install).

Supported file formats

• Audio: MP3, WAV, M4A, AAC, FLAC, OGG, WMA, AIFF
• Video: MP4, MOV, M4V, MKV, AVI, WEBM, WMV, FLV

For video, TranscribeX transcribes the audio track and keeps the video for playback next to the transcript.

Add a file

Use either method:

• Drag and drop. Drag one or more files anywhere onto the main window. If you’re in another sidebar section, TranscribeX switches to Files.
• Import card. On the home screen, click Import and choose one or more files.

Each file becomes a job in Files in the sidebar. If you add a single file, TranscribeX opens it right away. If you add several, they stay in the list and run one after another.

Auto-start or start yourself

Settings › Transcription › Auto-start Transcription is on by default, so a new job is queued straight away using your default model and language:

• Default Model: the model new jobs use. If it isn’t downloaded, TranscribeX uses the first downloaded model you can use.
• Default Target Language: the spoken language, or Auto Detect for models that support it.

If you turn auto-start off, new jobs wait as Pending. Open the job, check the settings below, and click Start.

Pick a model and language for one file

At the top of an open job, the bar has a model menu, a language menu and the start button.

1. Click the model menu and pick a downloaded model. Models are grouped by family (Nvidia, Whisper, Apple Speech and so on). More Models... opens the model settings to download others.
2. Click the language menu and pick the language spoken in the file. The list changes with the model, because each model supports different languages.
3. Click Start. To run a finished job again with other settings, change them and click Restart.

Not sure which model? See choose a model. Picking the right language instead of auto-detect usually gives better results.

Follow progress

While it works, the job shows a status badge such as Queued or Running and a progress bar. Click Cancel to stop. With Settings › General › Notifications on, you get a notification when it finishes, so you can switch to other apps.

If a job shows Failed, the error explains why, for example a Pro model used without Pro (“<model> requires Pro.”). Pick another model and click Restart.

Open the result

Click the job in Files to open it. The transcript appears in segments next to the audio or video player. From here you can:

• Play the recording alongside the transcript and fix any mistakes (see edit transcripts).
• Click Export in the toolbar to save or copy the transcript. Text is free; other formats need Pro (see export).
• Use the More options menu for Copy Transcripts, Show source file in Finder or Delete.

With Use AI to Auto-generate Title and Description on (in Settings › Transcription), the job also gets a title and description from your default AI service.

Next steps

• Many files at once, or a folder that fills up over time: bulk transcription (Pro).
• A video on YouTube or another site: download from URLs.
• Label who said what: speaker diarization (Pro).