Apple SpeechAnalyzer: The New King of On-Device Transcription

Apple's new SpeechAnalyzer API significantly outperforms both its predecessor and Whisper Small in accuracy and speed for on-device English transcription. While Whisper still leads in language support, SpeechAnalyzer is now the strongest option for English on iOS and macOS. Developers are encouraged to migrate to the new API to achieve a 4x reduction in word error rates compared to legacy tools.

Key Points

• Apple's new SpeechAnalyzer is 3.5x to 4x more accurate than the legacy SFSpeechRecognizer API.
• SpeechAnalyzer outperforms Whisper Small in both accuracy and speed, running roughly 3x faster on M2 Pro hardware.
• Whisper remains the preferred choice for non-English languages and cross-platform support, as Apple's engine currently supports only about 30 locales.
• The benchmark methodology was verified by reproducing OpenAI's own published Word Error Rate (WER) numbers for Whisper models.
• The developers of Inscribe have updated their app to prioritize SpeechAnalyzer for English transcription based on these results.

Sentiment

The overall sentiment is cautiously positive but highly qualified. Hacker News broadly agrees that SpeechAnalyzer is impressive and likely important for Apple users, yet the community resists the article's strongest framing because the benchmark does not cover enough models, languages, platforms, or real-world audio conditions. The thread is mostly constructive and technically detailed rather than hostile.

In Agreement

• SpeechAnalyzer appears to be a major leap over Apple's older speech recognition stack and a credible default for on-device English transcription on Apple hardware.
• Developers who have used the API in real apps report strong speed, accuracy, and live transcription behavior, with streaming results seen as a major user-experience advantage.
• Because the model is built into the operating system, apps can avoid bundling large models and users can benefit from a shared local engine across multiple transcription tools.
• The comparison to Whisper is still practically useful because many current transcription apps and local voice pipelines are built around Whisper-family engines.
• Local processing is appealing for privacy, latency, power efficiency, offline use, and accessibility scenarios where cloud transcription is undesirable or unavailable.

Opposed

• The benchmark is too narrow because it focuses on English read speech and does not evaluate noisy audio, meetings, accents, multilingual dictation, diarization, or domain-specific jargon.
• Whisper Small is not viewed as the strongest modern baseline, and commenters wanted comparisons against newer systems such as Parakeet, Nemotron, Voxtral, MOSS, Cohere Transcribe, Qwen ASR, and larger Whisper variants.
• Apple's limited language coverage and weak automatic language detection make it less compelling for users who regularly dictate across languages or need non-English transcription.
• A closed Apple-only model is not a replacement for open-weight or cross-platform systems, especially for Linux, Windows, Home Assistant, and workflows requiring inspectability.
• Some users remain skeptical because Apple's existing dictation and Siri experiences still struggle with accents, technical terms, context, and proper nouns in everyday use.