Apple Intelligence’s real-time transcription much faster than OpenAI’s Whisper

Apple Intelligence has been found to provide real-time transcription of audio into text at double the speed of Whisper, the equivalent technology from OpenAI.

iOS 26 translation in Messages.

MacStories writer John Voorhees along with his son wrote a simple command line tool, dubbed Yap, using Apple’s SpeechAnalyzer and SpeechTranscriber frameworks introduced with iOS 26, iPadOS 26 and macOS Tahoe 26 that enable developers to implement transcription features in their apps without breaking a sweat.

What they found was that Apple’s technology transcribes spoken word approximately two times faster than Whisper, OpenAI’s speech recognition system unveiled in September 2022 which understands accents and can distinguish between background noise, speech and technical jargon. It also significantly outperformed MacWhisper in terms of real-time translation.

Real-time transcription of audio into text via Apple Intelligence is much faster than OpenAI Whisper

Voorhees says Apple Intelligence-based Yap transcribed a seven-gigabyte 4K video file (about 34 minutes long) into subtitles in 45 seconds. Contrast this with the MacWhisper Mac app (built on OpenAI’s Whisper technology) which transcribed the same file in 3 minutes and 55 seconds using the Large C2 large language model (LLM). Switching to the Large V3 Turbo LLM enabled MacWhisper to finish the job significantly faster at one minute and 41 seconds, which is still more than two times slower than the Apple Intelligence counterpart.

Apple Intelligence uses Apple’s own LLMs for transcription. I’m not a big fan of Apple Intelligence, but live transcription and translation are some of the most impressive features on iOS 26, iPadOS 26 and macOS Tahoe 26.

As we noted in our WWDC25 roundup, live translation on iOS 26, iPadOS 26 and macOS Tahoe 26 is available in many places throughout the system. In the Phone app, the feature translates calls in foreign languages. When recording calls, a transcription of the call is automatically saved in Notes.

FaceTime video call on iPad with live translation
Live translation on FaceTime. Image: Christian Zibreg/iDB/Apple

On FaceTime video calls, you see live captions as participants talk. In the Messages app, your outgoing messages are automatically translated into a foreign language as you’re typing the message, and vice versa. In the Music app, Apple Intelligence powers on-the-fly translation of real-time lyrics.

iOS 26 translation in Messages.

As mentioned, Apple’s new speech frameworks are available for third-party developers to implement in their own apps. OpenAI’s Whisper has also been available for third-party development, so it already powers a bunch of AI apps, including MacWhisper that Voorhees used in his testing.