If you already have a meeting recording, interview, lecture, or video file, you do not need to play it aloud and dictate it again. A Mac transcription workflow can prepare the recording, run speech recognition, and give you a transcript with timestamps and speaker context for review.
1. Choose a local or remote processing path
Start by deciding where the model work should happen:
- Use a local ASR channel when you want the recording processed on your Mac.
- Use a remote ASR provider when you need a provider model or language capability and have configured your own key.
- Keep enhancement separate from ASR when you want to transcribe locally but summarize or rewrite with a different model.
The important distinction is not only “which transcription app,” but which part of the recording is allowed to leave the device. Check the provider policy before using a remote path for sensitive material.
2. Import the recording
In Voxt, choose an audio or video recording from the file workflow, or drag it into the import area. Video files can be analyzed when they contain a supported audio track. Voxt prepares the media and extracts the audio needed for transcription.
The file workflow is designed for recordings that already exist. For a live call or presentation, use Meeting or Captions mode instead so you can follow the transcript while the session is happening.
3. Let the file task finish
Longer recordings run through a queued analysis path. You can monitor progress, cancel work, prioritize another file, or retry a failed task. The current meeting-file workflow supports recordings up to 12 hours, while actual processing time depends on the file, model, Mac hardware, and audio quality.
4. Review speakers, timestamps, and decisions
Treat the first transcript as a working record, not an unquestionable source of truth. Review:
- Names and technical terms
- Speaker labels
- Timestamps around important decisions
- Places where multiple people speak at once
- The summary and action items before sharing them
When speaker analysis is available, the meeting result can make it easier to find who said what. You can then use the result for summaries, conversation analysis, and follow-up questions.
File transcription versus live dictation
File transcription starts with audio or video that already exists. Live dictation starts with your microphone and returns text to the app where your cursor is focused. Voxt supports both, but the workflows solve different problems:
- Use file analysis for meetings, interviews, classes, presentations, and saved recordings.
- Use hold-to-talk transcription for email, chat, notes, browsers, editors, and other daily writing.
- Use Meeting or Captions mode when you need a live transcript of system audio, microphone audio, or both.
That separation keeps the interface honest: a tool optimized for a saved video file does not automatically replace a system-wide writing shortcut, and a dictation tool does not need to pretend it is a full video editor.
A short privacy checklist
Before transcribing sensitive audio or video:
- Confirm whether the selected ASR and LLM channels are local or remote.
- Review the remote provider's retention and training policy if you use BYOK.
- Check where the local recording and transcript are stored on the Mac.
- Review speaker labels and summaries before sending the result to anyone else.
For Voxt's current data-path explanation, see Privacy & Data Flow. For the product workflow, see Local audio and video transcription for Mac.