Live Transcribe (by Mighty Fine Apps) for iOS/iPadOS, Fall 2025   

We now return with the latest installment of the Auto Captions project with a review of Live Transcribe by Mighty Fine Apps.

Bottom line up front: Great app, and the captioning tool I use most on a day to day basis. It depends on internet access to a cloud server for best results. As a cloud-based app, there are some costs and downsides inherent to how it works - running on other peoples' computers, I mean, the cloud, costs money. It also has advantages - you too can witness the firepower of a fully armed and operational data center doing audio processing. This makes it ridiculously resilient against background noise, and gives it the ability to lock onto the conversation you want.

I should start off by saying that as much as I love this app, I wish they'd named it literally anything else because Google has had their own captioning app with the same name. I start seemingly every conversation about it with some variation of "Live Transcribe, no, for iOS, not the Google one." On the other hand, I originally found this by searching for "live transcribe" in the App Store so... maybe their marketing worked out pretty well!

To save keystrokes, bytes, and my sanity, I'm referring to the application in this review solely as Live Transcribe.

What it can do

According to my App Store order history in GMail, I've used Live Transcribe for almost three years, in any and every scenario I can imagine short of air travel:

  • I've used it around the dinner table with my family.
  • I've used it in the car to chat with other people in the vehicle (to clarify: I was not driving).
  • I've used it via CarPlay so the microphones in back pick up what those passengers say.
  • I've used it in the car to caption both sides of a phone call held via CarPlay over the car's audio system (again: not driving).
  • I've used it at the doctor's office.
  • I've used it in the grocery store while talking to the staff.
  • I've used it for random conversations outside with the neighbors.
  • I've used it to talk to the service tech at the car dealership's noisy service counter.
  • I've used it for small family holiday gatherings to follow conversations around the room.
  • I've used it at large family gatherings for the same reason.
  • I've used it at technical conferences like Codemash and Stir Trek, actually understanding the presentation I'm attending.
  • I've used it (occasionally) to caption video calls through a combination of witchcraft and a TRRS cable.

I've never used it on an airplane, although in the basic offline mode, I could... though that would have downsides, which we'll get to. Also, bear in mind that 95% of my use of this app is on my iPhone - however, it does work on the iPad and I've used it quite a bit there, too. I just don't think there's anything particularly remarkable or different about the experience on an iPad.

A note on timing

I started writing this review in early September, 2025, and I've been collecting my notes on it for at least six months now. I'm aware Apple made major updates to their on-device live captioning recently with this year's OS updates, and I will have more to say about them soon. In the mean time, this app stands on its own merits.

What's good

It's difficult to go deep on all these scenarios; there's a lot of them. What impresses me most with Live Transcribe is what I alluded to above - it possesses a truly freakish ability to lock in on the specific conversation that you want it to, with little effort on your part. Sometimes when I'm talking with someone on my left and the conversation shifts to someone on my right, I'll slide my phone over, and it switches to prioritizing captioning that person instead. This works even when the person to my left continues talking to someone else. This helps in noisy environments like that car dealership service counter: the service tech helping me asked me to wait while they entered data into the computer. With captioning running, Live Transcribe started to caption the local side of the person a few places down along the counter talking on the phone. At some point while that conversation continued, the first tech started talking to me, and Live Transcribe immediately switched to captioning what they said to me even with someone nearby talking loudly on the phone. I don't know how they do it, and it's uncanny.

I attended two technology conferences in 2024: Codemash in January for the first time in over ten years, and Stir Trek in May after a few years' break. These were my first two conferences in what I now think of as the "post captioning" era - one in which assistive tools like automatic captions are ubiquitous, easily available, inexpensive, and work well. Codemash 2024 was the best technical conference experience I've had in my lifetime, solely due to Live Transcribe. Whether hallway chats with friends between sessions, mealtime conversations with my peers, actual presentations, or chats with the presenter afterward, it didn't matter. Everything was perfectly covered. Between the Wednesday night opening lightning talks and the conversations I mentioned, I easily burned 20+ hours of captioning time in two days. 100% worth it.

The best thing: for particularly challenging situations, I have a Hollyland Lark M1 wireless microphone system. This comes with excellent built-in noise cancellation and 200 meters of range, so it's great for a conference environment. Turned out, this was totally unnecessary. The conference center Codemash uses wires their rooms with speaker systems so I could just sit down, prop my phone on my leg, and get perfect captions without doing the "oh do you mind the extra microphone" dance. This helps at Codemash, particularly, because of a culture of "if the talk's not working for you, just go to another" and this way I needn't interrupt a speaker to ask for my microphone back.

I don't need to resort to weird hacks like external microphones as much as I used to, though. Unless it's an absurdly noisy environment or I need to caption audio from 100+ meters away[1], I can get away with just using my phone (currently, an iPhone 15 Pro Max) with its onboard microphone. This app is absurdly resilient against environmental noise. Take the scenario I mentioned above: captioning both sides of a call in the car over CarPlay. I don't mean that I was using my phone to make the call and drive CarPlay and just happened to take a call on it at the same time. I mean my wife took a call from a family member on her phone, played over the vehicle's speakers through CarPlay. I started Live Transcribe on my phone and it ignored all the road/wind noise and captioned what both my wife and my sister-in-law were saying.

Live Transcribe has another neat trick, though I actually don't use it frequently: speaker identification. It's not by name (although that'd be a neat trick), but rather by speaker diarization to detect switches between speakers. For my typical use cases I don't need this frequently, but it's nice to have. You activate it by starting the higher-quality captions option (the big "Transcribe" button on the main screen, as opposed to "Basic Captions"). At the top of the screen there is a little pill that says "Pro" and tapping on it lets you switch it to the beta "Ultra" captions, which will attempt to color-code the different speakers detected. Speaker diarization is a little bit of a black art, so it's not 100% accurate, but in a rapid-fire conversation it could be helpful for keeping track of when the speaker changed.

I particularly like how they handle pricing for the Ultra captions. Rather than require you to subscribe to an additional tier or purchase a separate block of hours, they charge Ultra-level captions against your minute allotment at a 1.5x rate. So - 10 minutes of Ultra captions would be charged as 15 minutes against your available time. I think it's a pretty elegant solution.

What's less good

Connectivity, unfortunately, is the weak link. Stir Trek 2024... was a somewhat different story than Codemash. Unlike the conference venue Codemash uses, Stir Trek holds their event at a movie theater. They lack the budget to set up extensive wifi for a one day event, and the theater hosting it was a black hole for cell signal in 2024[2]. This left me, in most cases, with no internet service, falling back solely to what I could use on-device. Unfortunately, this meant using Apple Live Captions (not great at the time), or falling back to Live Transcribe's "basic" captions mode which works offline (also uses Apple's onboard live captioning). There was also the option of using to a few alternate tools built on Whisper, but I don't love Whisper-based live captioning. It's a fantastic tool for captioning prerecorded audio[3], but every approach I've seen using it on live audio is a hack, doesn't yield reliably good results, and burns through your battery in no time.

Unfortunately, this is a fundamental problem in live captioning. If your approach depends on pushing a lot of bits up to the server so it can do processing magic on them, what happens when the server isn't accessible or working? To its credit, Live Transcribe does try to give you a sense of how well this is going to work. Once you bring up the captioning screen, it will try to evaluate your connection and tell you if it is one of: poor, average, or good. If you get average or good, everything should work fine. Poor, as you might expect, is a crap shoot. It will sometimes work, but you can't depend on it and it may end up buffering a lot of audio or running very noticeably slow.

It's difficult for me to regard this as a knock against Live Transcribe, specifically. They've made a great app and it's not their fault you need network access to work properly. It is a little frustrating when you need good captioning, on-device isn't cutting it, there's no wifi, and you're in a cellular black hole.

Another minor issue - and again, this is not necessarily a knock against Live Transcribe specifically - is that battery use can be heavy. Pushing all that audio data up to a remote server uses power.

If you have significant privacy concerns, an app that records everything your phone hears and pushes it up to a cloud server for processing might not be a thing you want.

Other notes

This app has been life-changing for me and that would be true even if it were ugly as sin and set my retinas on fire every time I used it. Is it going to win an Apple Design Award? Probably not, only because it's niche and not "cool" enough. But... it looks good, it has light and dark modes, it allows configuring font face and sizes.

Other things to note:

  • You can save or not save transcripts; the default setting is that if you X out of the captions display, it saves the transcript.
  • You can delete saved transcripts.
  • It has an option to show a button that uses AI to summarize the last minute of the conversation, or to summarize a transcript on save.
  • If you don't want the summarization on save, you can turn that off in Settings -> Advanced Settings.
  • If (as I do) you have a kid who likes to look over your shoulder at your captions, it has an option to censor profanity.

What it costs

A subscription to use the app is (so far as I can tell) mandatory. Subscriptions include a base level of captioning time per month. Subscription options:

As of the completion of this post (October 2025), here is what pricing looks like:

  • Essentials: ($10 monthly, $80 yearly ($6.67/mo)): 5 hours per month.
  • Daily: $29.99 monthly, $324 yearly ($27.08/mo): 20 hours per month.
  • Professional: $75 monthly, $810 yearly ($67.50/mo): 50 hours per month.
  • Unlimited: $150 monthly, $1600 yearly ($133.33/mo): Unlimited.

In addition to the subscription, you can buy pay-as-you-go hours (also as in-app purchases) for approximately $1.50 an hour. You can purchase in 1, 5, 10, 20, 50, or 100 hour blocks.

Subscription hours do not roll over; use them or lose them. Unused pay-as-you-go hours do roll over until used up. A customer-friendly feature: after your beginning-of-the-month rollover, the app uses your subscription hours preferentially before it starts using your additional paid hours. This is a small thing I really appreciate.

My verdict

I love this app. I use it weekly, if not daily. It not only lives on my home screen, I have the widget on my lock screen so I can drop straight into captions with a single tap. I appear to be grandfathered into a slightly cheaper version of the Essentials plan, and even with the improvements to on-device live captioning, I don't see myself dropping that subscription in the near future. Every 2-3 months I find myself throwing another 10 hours in. I'm planning on attending Codemash again this coming January, and no doubt I'll be throwing another 20 hours or so in for that.


  1. This has never happened for me. I think about 75 feet was my maximum distance. ↩︎

  2. Happily, the cellular companies improved service in the area for Stir Trek 2025. ↩︎

  3. Though, in my recent experience, Nvidia Parakeet v3 is equally as good as Whisper Large v3 Turbo while being ridiculously fast. ↩︎

View All Posts