iOS 27 beta 3 unlocks two live sliders over Siri’s voice: Pace for speed, Expressivity for emotional charge. The audio preview updates as you drag. The feature stays limited to iPhone 17 Pro and iPhone Air, the only devices able to run the rendering fully on-device.
Key Takeaways
- Two sliders, Pace and Expressivity, activated in iOS 27 beta 3 released on July 6.
- The audio preview updates in real time as soon as either slider moves.
- Limited to iPhone 17 Pro and iPhone Air, the feature runs entirely on-device.
Have an AI Sum Up This Article
ChatGPTPace and Expressivity, with instant preview
Apple had slipped the Pace and Expressivity labels into the earliest iOS 27 betas, tagged “Coming soon”. Beta 3 turns them on for real. Users first pick a voice from the expanded catalog, then fine-tune with two sliders.
The Pace slider changes speech speed. The Expressivity slider adjusts modulation, intonation and emotional weight. As soon as a slider moves, Siri plays a rolling audio sample that reflects every tweak in real time. No reload, no lag.
The setting propagates beyond Siri. Apple Maps and Safari switch to the same voice for turn-by-turn navigation and reader mode. For the first time, a user who prefers a fast, neutral voice on navigation gets the exact same timbre when Siri reads a message aloud.
Three apps sharing the same voice suggests a unified voice infrastructure behind Siri, Maps and Safari. It is no longer a per-app TTS engine, but a shared layer that Apple drives end to end.
On-device is a premium buying criterion again
The feature only turns on for iPhone 17 Pro and iPhone Air. The stated reason is technical: expressive voice rendering runs locally on the on-board Neural Engine. Every previous model, including the standard 17, does not receive the feature.
This is a strong signal of the Apple Intelligence strategy. After a year spent splitting capabilities between local and Private Cloud, Apple pulls voice features back to pure on-device. Perceived latency becomes a commercial argument, and the Pro tier price justifies the dedicated silicon.
The competitive backdrop pushes the same direction. Google promotes Gemini Live on Pixel with expressive cloud-side voice, Samsung leans on Galaxy S26 to distribute generative features. The Apple push on iOS 27 now runs through the voice showcase.
For developers, the consequence is immediate. A voice chatbot embedded inside an iOS app will not carry the same personality depending on the model running it. The hardware segmentation of AI experiences deepens, and UX guidelines will need to reflect that.
Also on Horizon:
- Zhipu Ships China’s Free Rival to Claude With ZCode
- Claude Code Ports Command & Conquer iOS in 40 Minutes
- Alpha School Sells $75K AI Schools to Wealthy US Families
Voice becomes the new OS battleground
What is at stake with Pace and Expressivity reaches past Siri. It is voice returning as an OS differentiation surface. After fifteen years of voice assistants doing little more than reading reminders, Apple treats audio with the same care as typography or animations.
In the short term, the change mostly hits passive use. People who listen to long articles while walking, or use Siri to read messages while driving, finally get real control. It is a premium accessibility case as much as a comfort feature.
Over three to six months, the story to watch is the broader Siri overhaul already started with help from Gemini. A personalizable voice takes its full weight if Siri itself becomes conversational and able to hold a thread across multiple turns.
The competitive counter-move is predictable. Google will most likely ship equivalent sliders on Gemini Live by year end. The difference will hinge on perceived latency, realism and whether the sliders actually change how the user feels the voice. A gimmick slider changes nothing, a slider that reshapes perception does.
One open question remains: cloned voices. Apple still does not let users generate their own voice, unlike ElevenLabs or consumer clone tools. That is probably the next step, provided Apple clears the fraud and biometric consent questions upstream.
Follow the story on Horizon.
