Skip to content
Back to the lab
PunkFox

Bringing Next-Gen Speech-to-Text to Your Pocket: LiveSTT Now Features Apple Speech Analyzer & SenseVoice-Small

LiveSTT now features iOS 26’s Apple Speech Analyzer for instant, local STT, plus lightning-fast AI transcription via SenseVoice-Small.

changelog STT changelog

Bringing Next-Gen Speech-to-Text to Your Pocket: LiveSTT Now Features Apple Speech Analyzer & SenseVoice-Small

We are thrilled to announce a major update to LiveSTT! Our goal has always been to provide the fastest, most private, and highly accurate speech-to-text experience on iOS. With this latest release, we are taking a massive leap forward by integrating Apple’s brand-new native speech framework alongside our advanced AI models.

You can now experience the core features of LiveSTT instantly—with zero initial downloads required.

Meet Apple Speech Analyzer: The iOS 26 Transcription Breakthrough

Introduced at WWDC 2025 for iOS 26 and macOS 26, the Apple Speech Analyzer is Apple’s cutting-edge Speech framework API. By integrating this natively into LiveSTT, you can now enjoy high-fidelity transcription right out of the box, utilizing approximately 40 languages and regional variants through SpeechTranscriber.

A Modular Architecture for Smarter Audio

The beauty of Apple Speech Analyzer lies in its modular design. Instead of a one-size-fits-all approach, it utilizes a central manager (SpeechAnalyzer) that coordinates specific sub-modules:

  • SpeechTranscriber: Handles high-performance, ultra-accurate speech-to-text.
  • SpeechDetector: A built-in Voice Activity Detector (VAD) that intelligently senses exactly when someone is speaking, optimizing processing power.

Why It Changes the Game for LiveSTT Users

  • 100% On-Device Privacy: Your audio never leaves your iPhone. All processing happens locally, ensuring absolute data privacy and instant execution.
  • Built for Long-Form & Far-Field Audio: Unlike previous Apple tools, this framework is engineered specifically for long conversations, lectures, and meetings, even when the speaker is far from the microphone.
  • Seamless Live Transcription: It taps into Swift’s modern concurrency features (AsyncSequence) to stream text live as you speak.
  • Apple-Grade Reliability: This is the exact same underlying technology power-housing Apple’s native Notes, Voice Memos, and Journal apps.

The Evolution: Legacy API vs. The New Speech Analyzer

If you’ve used older transcription apps, you might remember the limitations of Apple’s original speech framework. Here is how the new native engine stacks up against the legacy version:

FeatureSFSpeechRecognizer (Legacy)SpeechAnalyzer (New)
IntroductioniOS 10iOS 26
Best Used ForShort text dictationLong-form audio & real-time streams
PerformanceStandard latencySignificantly faster execution
FlexibilityLow (rigid framework)High (Modular design with VAD)
ProcessingOften required server assistancePurely on-device

Powering Advanced AI: Continued Support for SenseVoice-Small

While Apple Speech Analyzer provides an incredible, download-free baseline, LiveSTT remains committed to offering elite-tier AI options. For users who need deeper audio insights, we continue to fully support SenseVoice-Small by FunAudioLLM.

What makes SenseVoice-Small special? It is a non-autoregressive end-to-end AI model designed not just for text conversion, but for multi-lingual accuracy, emotion recognition, and audio event detection (like clapping, laughter, or background noise).

Unparalleled Inference Speed

SenseVoice-Small delivers massive performance gains while keeping its footprint tiny:

  • 5x Faster inference speed than Whisper-Small.
  • 15x Faster inference speed than Whisper-Large.
  • Consistent Latency: Unlike older models, SenseVoice’s processing time does not spike drastically as your audio file gets longer.
  • Mobile Optimized: Compressed down to just 448MB, it runs flawlessly on modern iPhones without draining your battery.
  • Language Specialization: Exceptional localized support for English, Chinese (Mandarin), Cantonese, Japanese, and Korean.

Choose the Way You Transcribe

With this update, LiveSTT gives you the best of both worlds. Use the native Apple Speech Analyzer for an instant, lightweight, and completely local setup covering 40 languages. Switch over to SenseVoice-Small when you need advanced multi-lingual AI capabilities, ultra-fast file processing, and rich environmental audio detection.

Update your LiveSTT app from the App Store today and experience the future of mobile on-device transcription!