Is Apple’s new free AI-powered transcription feature a game-changer for journalists? We compared it to paid services

If you’re an iPhone user who doesn’t delay updates until your phone is on the brink of breakdown, you may have noticed something new in your Notes app. The iOS 18.1 operating system, which hit our pockets this week, included a new feature for phone calls and the Voice Memo and Notes apps that take talk-to-text to a new level — allowing users to record and transcribe entire audio interviews on both mobile and desktop. The new abilities also represent Apple’s foray into a new area: the transcriptions are accompanied by an AI-generated summary. 

Sounds amazing, right? We’re as excited as any journalist who never wants to hand transcribe another interview could be, but we’d be doing the profession a disservice if we didn’t ask questions, obviously. 

First up, it only works on desktop models running the latest version of the Sequoia system software and the AI features can only be used on iPhone 15 Pro, iPhone 15 Pro Max and the entire iPhone 16 series. In addition, according to Apple, “audio transcription is available only in English for the variants of Australia, Canada, Ireland, New Zealand, South Africa, UK, and US.” We called to clarify, and the rep couldn’t point to any specific list of supported languages, saying only that, “supported languages will automatically transcribe.” So, I guess you just have to give it a whirl to see if it’ll support your language needs.

Language functionality aside, this is exciting news, especially when you consider that other top speech-to-audio transcription services, Trint for instance, can run users as much as $50 per seat per month, and that’s just for the basic package. So, could this new free software be the game changer we hope it is? We wanted to see for ourselves, so we put Apple’s new functionality to the test against some of the top paid services being used by journalists today, namely, Otter, Zoom, Trint, and Sonix. We ran an audio interview through all of them and compared the output to a cleaned-up version to see who got what right.

We found that, as much as we hate to say it, you do get what you pay for. But that’s not to say that the new Apple transcription feature isn’t something to celebrate. At the end of the day, it comes down to needs, if you’re looking for a simple transcription service to serve as a backup to informal interviews conducted in low-stakes environments without much background noise – Apple has your back. But if you’re looking to transcribe during a meeting or interview with numerous voices and/or in an environment with lots of background audio, or to convert your interview from an existing video or other media file, you may want to consider one the pricier competitors.  

Name	Functionality	Accuracy	Editing functionality	Cost	Speaker differentiation
TRINT	An AI-based software that transcribes video clips and other media files.	Our analysis needs to be caveated with the fact that Trint cuts us off before the interview was finished, given that we were only using the free trial. That said, the portion of the interview we did receive was noticeably cleaner, both in terms of speaker differentiation and general dictation, than the other platforms, which is perhaps unsurprising given that Trint is by far the costliest of the lot.	Functionality is more expansive than most other options we found, with an interface that allows you to combine transcripts into a script, rough cut, or edit with timecodes, etc.	Free trial that gives very limited functionality, then $52 per seat per month/$60 per seat per month for advanced content creators / teams with automation.
OTTER	As standalone app, Otter.ai also integrates into platforms like Google Meet and Zoom. It is the technology powering Zoom’s virtual meeting transcription, “joining” your meetings as another attendee. The software uses machine learning algorithms to process audio recordings into structured text documents.	Otter had a noticeably hard time differentiating between the speakers in our transcribed interview. At times, it nailed it, but other times it struggled to understand when one person stopped and another started talking, which resulted in the interview requiring fine-tooth comb editing thereafter.	Allows you to stitch a transcript to an audio file to give you a time-stamped conversation.	A free plan that caps you at 300 monthly transcription minutes / 30 minutes per conversation. Paid plans start at $8.33 per month & caps you at 1,200 monthly transcription minutes and 90 minutes per conversation.	Otter states that it does distinguish between speakers, but the differentiation is noticeably sensitive to other interfering audio.
APPLE NOTES	In the Notes app, software can record audio and have the spoken words transcribed to text.	The most glaring issue is the software’s inability to differentiate between speakers, which rendered the cleanup process for the transcript far more labor-intensive process than on the other platforms. That aside, there were also other repeated mistakes that the other platforms caught, mistaking “note taking” for “notaking” for instance among many others.	Functionality also icludes searching the transcript, adding text, and copying the text to other documents.	Free	Does not distinguish between speakers.
Sonix	Sonix offers AI transcription and time-code stitching for media files. Similar to Trint, it’s designed to generate transcripts for audio and video files like podcasts and lectures or interviews. The platform also includes helpful features like background noise reduction, a big one for anyone who’s had to take an interview in a noisy restaurant.	Overall, Sonix held its own in this comparison in that it didn’t miss key phrases or entire pieces of audio. That said its transcript was not without error. There were notable issues with its ability to differentiate between speakers, as well as a few instances of repeated words. Finally, the system repeatedly mis transcribed a single word, “chronolocation,” however given how highly technical that word is, it’s no surprise that all the platforms we tried had the same issue.		Sonix has a free trial, then paid plans start at $10 per hour, which gives you pay-as-you-go AI transcription and 10GB of storage for 90 days.	Does differentiate between speakers
Zoom	According to its website, Zoom is a little different because it relies on integrated transcription services such as Otter.ai Tactiq, Grain, Krisp & Happy Scribe among others. So, its functionality is a result of the third-party employed.	The platform was able to differentiate between speakers, perhaps unsurprisingly as the audio was pulled from each speaker’s separate computer, but the transcript otherwise featured typos and misunderstood phrases resulting in the need for careful proofreading thereafter.		Most of us get Zoom through our employer or educational education, but for those who don’t you’ll need to upgrade to at least the pro version ($13.32/month/user) if you want the AI accompaniments included – transcription among them.

Name	Allow for real time transcriptions	Support different languages/dialects	Transcribes video & audio	Share live transcription feed with colleagues for real-time collaboration	Automatic language detection on live transcription for multi-lingual events
TRINT					It can transcribe voice recordings and live speech in over 40 languages. According to the website, the live transcription app can automatically detect which language is being spoken and switch between different languages as it transcribes.
OTTER					it can only do one language at a time
APPLE NOTES					Unclear, but as of now it only promise to be compatible with English.
Sonix		According to the website, Sonix can turn your audio and video-to-text files in 40+ languages.
Zoom

The post Is Apple’s new free AI-powered transcription feature a game-changer for journalists? We compared it to paid services first appeared on Storybench.