Amlistening vs MacWhisper, Vibe, Wispr Flow and OpenAI
Each of these five is good at a different job. Choose by what you need.
Amlistening is an offline app for spoken observations in the field. It keeps each word as you said it, and leaves summaries and clean-up to your own AI agent, on purpose.
- Published
- Competitor facts checked
| Feature | Amlistening | MacWhisper | Vibe | Wispr Flow | OpenAI’s API |
|---|---|---|---|---|---|
| Where it runs and what it records | |||||
| Transcribes on your own machine | Yes | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | No. Source: Wispr Flow’s data controls | No. Source: OpenAI’s realtime transcription guide |
| Shows the transcript while you speak | Yes | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | Yes. Source: wisprflow.ai | Yes. Source: OpenAI’s realtime transcription guide |
| Transcribes files you already have | Yes | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | No. Source: Wispr Flow’s help on audio files | Yes. Source: OpenAI’s speech to text guide |
| Records meetings or system audio | No | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | Yes. Source: Wispr Flow’s pricing | Not applicable |
| Types into other apps | No | Yes. Source: macwhisper.com | Not stated | Yes. Source: wisprflow.ai | Not applicable |
| Publishes its source code | No | Not stated | Yes. Source: Vibe’s licence | Not stated | Not stated |
| While you record | |||||
| Starts and stops by a global key or by voice | Yes | Not stated | Not stated | Yes. Source: Wispr Flow’s help on hands-free | Not applicable |
| Controls a recording from the menu bar or tray | Yes | Not stated | Not stated | Not stated | Not applicable |
| Keeps the recording through a crash | Yes | Not stated | Not stated | Not stated | Not applicable |
| Warns when the microphone is silent | Yes | Not stated | Not stated | Not stated | Not applicable |
| What you get | |||||
| Leaves your words as spoken | Yes | Not stated | Not stated | No. Source: wisprflow.ai | Not stated |
| Marks your keywords where they are said | Yes | Not stated | Not stated | Not stated | Not stated |
| Counts how often each keyword is said | Yes | Not stated | Not stated | Not stated | Not stated |
| Plays one line’s own audio | Yes | Not stated | Not stated | Not stated | Not applicable |
| Gives each line the place where it was said | Yes | Not stated | Not stated | Not stated | Not stated |
| Lets you correct the transcript | Yes | Yes. Source: macwhisper.com | Not stated | Not stated | Not applicable |
| Exports the transcript as plain text | Yes | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | Not stated | Yes. Source: OpenAI’s speech to text guide |
| Labels speakers | No | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | Yes. Source: wisprflow.ai | Yes. Source: OpenAI’s speech to text guide |
| Writes summaries | No | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | Yes. Source: wisprflow.ai | Not stated |
| Translates | No | Yes. Source: macwhisper.com | Yes. Source: Vibe’s README | Not stated | Yes. Source: OpenAI’s pricing |
- Yes
- No
- Its own page does not say
- An API is not an app
Read from each product’s own pages on 12 October 2026. Each mark about another product links its source. We did not use the other products and are not affiliated with them. If a mark is wrong, write to support@amlistening.app.
Amlistening: what you said, as you said it
It records your voice and transcribes it on your computer, with no network. Each recording is called a take. Nothing you record is uploaded: the security page lists what the app does contact.
No summary and no clean-up, on purpose
That job is for the AI agent you already use. Export the take as plain text, and it gets each word as you said it. In a report, that matters: you can check each line against its own audio, and a hesitation is data.
- The transcript appears as you speak.
- Your keywords are highlighted in the line, marked on the waveform and counted. Change the library, and earlier takes are matched again.
- A key or a spoken phrase starts a take, stops it and marks a moment, with the window hidden.
- A crash costs seconds, not the take. Try it: force the app to quit during a take, then open it again.
- A silent microphone shows an amber warning after 3 seconds, and the take keeps recording.
- Any line plays its own audio. Click its time.
- Import a GPS trace, and each line gets the place where you said it. The map draws with no map tiles, so where you were stays on your computer.
- A correction keeps the machine’s text beside yours, and Revert puts it back.
- Import a WAV, MP3, M4A or MP4 file, and it becomes a take with a transcript.
- Hesitations such as “um”, “uh”, “like” and “you know” stay in the transcript.
Where it does less: it does not label speakers or translate. It does not record calls or system audio, or type into other apps. It does not track where you are: a place comes from a trace that you import. It transcribes about 80 languages with the default model. On a Mac it needs Apple silicon. The source is not published.
What it costs: Free has no expiry and keeps 10 takes of 5 minutes each. Pro lifts the caps and imports traces. It is $7 a month or $60 a year, with no one-time price. It is not sold in the EU, the EEA, the UK or Gibraltar. Cancel any time, and the app falls back to Free with nothing locked.
Choose Amlistening if you speak your observations while you work, often with your hands busy. You need each word as spoken, your keywords marked, and no upload. If you carry a GPS logger, it also shows where you said each line. It runs on Mac, Windows and Linux.
When another one fits better
Vibe Free and open source
Choose Vibe if you transcribe audio and video files that you already have, and you want a free, open-source app with speaker labels and summaries. Its README promises “fully offline transcription, no data ever leaves your device”, on macOS, Windows and Linux. The source is under the MIT licence.
MacWhisper Files, meetings and dictation on a Mac
Choose MacWhisper if you work on a Mac, and you want one app for files, meetings and dictation that you pay for once. Pro is “Pay Once”, at €64 for a licence. After about a year, that has cost less than Amlistening Pro, and it stays that way.
Wispr Flow Speech turned into polished writing
Choose Wispr Flow if you want to write by voice in other apps, and you want text that reads as written, not as spoken. It “automatically removes filler words, adds punctuation, and formats your writing”. Its data controls page says “Transcription always occurs on the cloud.”
OpenAI’s API Transcription for the app you build
Choose an API if you build your own app, and your audio can go to a hosted service. OpenAI’s realtime transcription is priced by the minute: gpt-live-transcribe costs “$0.017 / minute”. ElevenLabs Scribe is a service of the same kind.
Try it on your own voice.
The download is free and needs no account.