Speech
- CohereDefault
- Parakeet v2Mac · English
- Voxtral-Mini 3BMultilingual
- Whisper Large v3 Turbo
- Whisper Large v3
- speakrsSpeaker turns
Transcription that never sleeps.
Otter and Notta upload the recording to make a transcript. Augarr runs the models on the machine. The app is on Mac and Windows.
Get Augarr$500 once. Lifetime. No subscription.
mp3, wav, flac, ogg
Mac and Windows
-
1.0
The file stays put.
Point it at mp3, wav, flac, or ogg. The file can sit on the computer, an external drive, or a network volume. Leave a folder of calls and walk away. It takes the next file when the last one finishes. The audio is not uploaded to produce the transcript.
-
2.0
Speech runs where the file is.
The speech model downloads once, then reads the recording on that machine. Parakeet v2, the English model, runs on Mac. Cohere, Voxtral, and Whisper run on Mac and Windows. On Windows, Cohere uses the GGUF build.
-
3.0
The same pass marks who spoke.
speakrs writes the turns, and when they happened. It runs on Mac and Windows.
-
4.0
Then it pulls the fields.
Extraction reads the transcript already on the machine. Qwen fills three groups. You can also ask the transcript a question with that same local model. A cloud model can be turned on in settings. It stays off until you do, and only then does the transcript leave the machine.
- Call overview
- Company, agent, and outcome, each pinned to the moment it was said.
- Customer
- Name and phone, taken from the call.
- Payments
- Card fields from the payment talk, then a Luhn check on the number.
The models.
Each name is listed once. A second quant or another gateway is not another model. Parakeet v2 is the Mac-only row. The other local models run on Mac and Windows. A cloud model stays off until you turn it on.
On the machine
- Qwen 3.5 4B
- Qwen 3.5 9B
Cloud, off until you turn it on
- MiniMax M3
- DeepSeek V4.1 Flash
- Qwen3.8 Flash
- MiMo V2.6 Flash
- MiMo V2.5
- MiMo V2.5 Pro
- GLM-5.2
- GLM-5.3 Flash
- GPT 5.6 Luna
- Hy3
- LongCat 2.0
- Union Alpha
Before you buy.
Does the audio leave the machine?
Not for the transcript. The speech models run locally. Extraction and questions stay on the machine when the model is local.
What do I run it on?
A Mac or a Windows PC. The models download once. After that, transcription runs on that machine. Parakeet v2 is Mac-only.
Will it work through a pile of calls?
Leave the files in the queue. It keeps taking the next one.
How much is it?
$500 once. Lifetime. There is no subscription and nothing to renew.
Where is checkout?
On Whop. One payment of $500.
Get Augarr.
$500 once. Lifetime. No subscription.
One payment on Whop. It stays yours.
Get AugarrThe recording stays on the machine.