Apple Intelligence offers a bunch of interesting features, but if you’ve tried most of them for more than a few minutes, you realise they’re not quite up to scratch compared to the best artificial intelligence (AI) tools.
Image Playground is fine, for instance, but not particularly useful. Writing Tools work well enough but aren’t as ground-breaking as Apple might make out. And the less that’s said about Siri, the better.
The idea that Apple Intelligence is falling behind is well illustrated by the way it works in Apple’s Notes app. Here, Apple Intelligence can transcribe your audio recordings. That sounds great — it could be useful for making notes on your lectures or remembering what was said in a key business meeting. As a journalist, I’d like to use it for transcribing interviews.
Except the problem is that Apple Intelligence doesn’t recognize speakers, so it requires tons of extra work to sort out who said what — especially in a lengthy recording. Even just a 30-minute audio track can require me to spend over an hour editing my notes to specify which phrase was uttered by which person. It’s a bore and a chore.
Coming back for more
Now, a different AI tool has come up with a great solution. That app is called MacWhisper, and it takes the Whisper transcription tech made by OpenAI, then wraps it up into a handy Mac app.
Importantly, the latest version of MacWhisper adds speaker recognition to its arsenal. As long as you use an AI model that includes speaker recognition, different speakers now have their names highlighted in individual colors, and their words are also divided into separate color-coded blocks for quick identification. MacWhisper does all this automatically, without you needing to do a thing.
It’s exactly what I wish Apple Intelligence was capable of. Because Apple Intelligence is built into macOS for free, I’d rather simply use that rather than a paid alternative. But with Apple dragging its heels on whipping its AI platform into shape, MacWhisper is filling an important hole in my productivity workflow.
It’s not the first time I’ve tried MacWhisper. I took it for a spin last year when I first came across it, but it fell by the wayside when I realized it couldn’t pick out individual speakers from my recordings. Now, speaker recognition — which MacWhisper’s developer says was its most requested feature — is drawing me back in.
Better than the alternatives
Before MacWhisper’s latest update, and with Apple Intelligence leaving me unimpressed, I’ve previously been using Otter. This offers fantastic quality when it comes to audio transcription, and it has had speaker recognition for years. But it also requires a subscription, and its monthly usage limits are quite strict for my uses. That means I’ve been searching for an alternative for a while.
MacWhisper’s speaker recognition requires you to pay, but it’s a one-off fee, which I prefer (and right now, you can get 15% off with the code FINALLY to coincide with the update). Because it uses OpenAI’s tech, the quality is superb. And for me, it’s better to pay a one-off fee and use the functionality as much as you like than to pay a recurring subscription and still hit the app’s limits.
Simply put, MacWhisper is showing Apple how it’s done. I expect that Apple Intelligence’s transcription abilities will improve with time, but given how much Apple’s platform is lacking in other areas, transcription might not be much of a priority for the company.
That means that right now, MacWhisper is set to become my transcription app of choice. If you record and transcribe meetings, lectures, audio messages and more, it’s worth a try.
Read the full article here
