Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor
Apple's new SpeechAnalyzer API, introduced with iOS 26 and macOS 26, achieves 2.12% word error rate on LibriSpeech clean speech, outperforming Whisper Small (3.74%) and its predecessor SFSpeechRecognizer (9.02%). It runs roughly three times faster than Whisper Small. Developers migrating from the legacy API can expect a 3.5-4x reduction in word error rate.
Apple's new SpeechAnalyzer API offers a clear accuracy and speed upgrade over both Whisper and the legacy SFSpeechRecognizer.


