pennyshout.A GUIDE FOR THE JOB AT HAND
For interviews and internal recordings, start with local transcription and subtitle export. These starting points reflect documented workflow fit.
WHAT MAKES THE CUT
Each tool must fit the task. This is a shortlist, not a universal quality ranking.
Local, open source
A cross-platform starting point with batch transcription and subtitle timing controls.
Keep cloud summary features off when the recording must stay on-device.
Local, open source
A useful fit for watch folders and command-line subtitle workflows.
The current Mac build needs Apple silicon; download a model before offline use.
Mac-focused workflow
Consider this for a Mac-focused recording and transcript-editing workflow.
Check edition-specific export capabilities and the paid upgrade.
THE PRACTICAL DIFFERENCES
| Tool | Published price | Processing | Key limitation |
|---|---|---|---|
| Vibe | Free Open-source app. Optional cloud services can have their own charges. | Local transcription; optional cloud features | Keep cloud summary features off when the recording must stay on-device. |
| Buzz | Free Open-source app. Optional remote API usage is billed by its provider. | Local models; optional remote API | The current Mac build needs Apple silicon; download a model before offline use. |
| MacWhisper | Free + Pro Free entry and a paid upgrade are offered. Check the official checkout for regional pricing and edition limits; we have not verified a universal Pro price. | Local models; optional cloud integrations | Check edition-specific export capabilities and the paid upgrade. |
Submagic and Kapwing are cloud workflows and do not meet the local-processing requirement.
Editorial examples are unpaid; any paid inclusions are labeled. We compare official published facts; we have not run an accuracy benchmark. Read our method.
Installation and model downloads can require internet. The requirement here is local transcription after setup.
We have not run a controlled Chinese-audio benchmark. Language support does not establish accuracy. Try the same representative sample before choosing.
See what a useful listing could look like.
