Here's why I switched from Otter to this clean audio transcription app

If you spend a lot of time in online meetings or virtual briefings, you’ve probably used Otter. The app used to be my go-to for transcribing interviews, meetings, and voice notes. One of the originals of AI-enhanced productivity tools, Otter was quick, mostly accurate, and easy to use. But like many good services, the paywalls kept creeping in. The free tier started to feel too limited, and I wasn’t keen on paying just to unlock basic functionality. On top of that, I wasn’t comfortable uploading sensitive audio to the cloud anymore.

So I started looking for alternatives that gave me more freedom and control. That’s when I came across aTrain — a self-hosted transcription app built around OpenAI’s Whisper model. It’s open source, runs locally, and gives me full control over my files and how they’re processed. I’ve been using it for a few weeks now, and honestly, I wish I’d made the switch sooner.

Setup that takes minutes, not hours

Skip the terminal, start transcribing

I’m no stranger to self-hosting, but if there’s an executable file available, I’ll always take that route over fiddling with Docker or the terminal. One of the things that pleasantly surprised me about aTrain was how easy it was to get started. I didn’t have to mess with Python environments or manually install dependencies. The developer provides a ready-to-run executable — just download it, launch it, and you’re in.

The app runs in a browser environment locally, without needing an internet connection. No terminal windows, no Docker containers, no backend setup. I dropped it into a folder on my desktop, double-clicked, and seconds later, I had a clean interface waiting for my first audio file. That kind of simplicity makes a big difference when you use a tool regularly.

aTrain also ships with Whisper’s large-v3-turbo model built in, so it works out of the box. You can choose to download other models too. While the tiny model may not deliver the best results, the large-v3 model is solid even for complex transcription tasks. For testing, I stuck with the default.

It just fits into my workflow

No limits, no compromises

Otter worked great, but I often found myself working around its limitations. Upload caps, file type restrictions, and missing features on the free plan meant changing how I worked just to fit its rules. aTrain doesn’t get in the way like that. It supports practically every audio and video format I’ve thrown at it.

You drop in your audio file, pick a model, and hit transcribe. Alongside, the app shows live progress and outputs clean, timestamped text. You can save it, copy it or edit it in your text editor of choice. No logins, no uploads, no upgrade nags. Just straight-up transcription. As simple as that.

It’s already become my go-to for a bunch of use cases. I’ve transcribed voice notes I recorded on my phone during commutes, pulled quotes after interviews, and even processed old recordings I never got around to transcribing because it felt like too much effort. Now, I just drop them into aTrain and move on.

And because everything stays local, I don’t have to think twice about the type of audio I’m working with. Client calls, personal recordings, NDA-bound press briefings — all of it stays on my machine. Otter can't match that peace of mind.

I tested it primarily on my M3 MacBook Air. On average, transcription took about 1.5 to 2 times the duration of the recording. That might be slow for really long files, but it’s a fair tradeoff for everything else you get. If you’re using a system with an Nvidia GPU, you can speed things up dramatically with CUDA processing. Transcription quality is impressive too, even in multi-speaker mode. In my testing, I found that even when not perfect, it got most of the way there. In fact, in most of my testing, it matched or beat Otter’s accuracy — and that’s from a commercial product built solely for transcription.

Lightweight by design, and that’s a good thing

One job, done right

aTrain doesn’t try to do too much, and that’s exactly why it works so well. You won’t find automatic summaries, collaboration features, or meeting tool integrations. And I’m okay with that. Instead, you get what matters: fast, accurate transcripts under your full control. It does one thing and does it well.

If you’re technically inclined, you can customize it further. I haven’t gone down that road myself, but since it’s open source, you can dig into the code, hook it up to other tools, or tweak the output pipeline. That said, the out-of-the-box experience is already great, and you don’t need to write a single line of code to use it.

It’s time to say goodbye, Otter

aTrain won’t be for everyone. If you depend on live collaboration, cloud syncing, or polished dashboards, Otter or one of its competitors might still be a better fit. But if you want privacy, simplicity, and full control without sacrificing quality, this app absolutely nails it. I started looking for a Whisper front end as a stopgap. What I found was a dead-simple, efficient tool I now use every week. And honestly, I haven’t missed Otter once.