Technology

iOS 27 Beta 3 Brings Voice Customization to Siri — Here's What Actually Changed

Martin HollowayPublished 4w ago4 min readBased on 5 sources
Reading level
iOS 27 Beta 3 Brings Voice Customization to Siri — Here's What Actually Changed

Apple released iOS 27 developer beta 3 on July 6, 2026, with a functional update to Siri's voice controls. Two new sliders—Pace and Expressivity—are now live after being visible but locked in earlier betas.

Pace controls how fast or slow Siri speaks. Expressivity controls how much emotional variation the voice carries, making it sound more human-like or more neutral. As you drag either slider, Siri speaks a sample phrase in real time ("You have one new message"), so you hear the change immediately without having to leave the menu and trigger Siri again. That live preview loop is a deliberate UX choice that cuts out the friction of commit-and-test cycles.

Apple announced these features at WWDC 26 in June, as part of a broader overhaul that positions Siri as an AI-native assistant. iOS 27 adds multiple ways to invoke Siri: voice trigger, side-button press, swipe down from the Dynamic Island, text input, or a dedicated standalone app. Beta testers also have access to a range of voices across different accents, all selectable in the same settings panel.

What These Controls Actually Do

Pace and Expressivity aren't new to speech synthesis technology—text-to-speech systems have let developers control speech speed and prosody (the emotional patterns in speech) for years, usually through APIs. What Apple is doing is exposing those controls to regular users, directly in the settings menu, without needing an app or workaround. That matters for people with auditory processing differences, non-native speakers, and anyone who finds the default cadence uncomfortable during longer conversations.

Expressivity is the more technically interesting control. Prosody in neural text-to-speech systems isn't just about tempo—it typically involves conditioning the speech model on emotional embeddings (mathematical representations of tone and feeling). A slider that adjusts this in real time suggests the voice model, running on-device or through a fast server connection, accepts that input continuously rather than generating a fixed style once. Apple hasn't published technical details, but the live-preview behavior indicates the inference is fast enough to feel synchronous.

Known Issues to Watch

TechCrunch reported that some users on X said they lost access to the new AI-powered Siri after installing beta 3, while others saw their device re-indexing data after the update. Both are expected behaviors for a beta that makes changes to how on-device models are stored and indexed.

Apple's release notes flag something worth attention: a known issue where disabling Siri may not reliably delete stored Siri and Dictation interaction data. It's a bug, not intended behavior, but it's the type of bug worth tracking through later betas given regulatory scrutiny on voice assistant data retention across multiple jurisdictions.

Beta 3 also refreshes the Reminders app icon. Icon updates in late betas sometimes signal a broader design refresh before the final release, so it's worth noting.

The Wider iOS 27 Context

The AI-powered Siri in iOS 27 is architecturally different from its predecessor. Apple's developer documentation describes Siri's language understanding as something that "evolves" in this release—a careful phrase that signals a model that updates or adapts between OS versions rather than staying static. The new activation methods and voice customization all point to a back-end rebuild that started at least as far back as WWDC 25.

For developers, the practical item is the release notes on developer.apple.com. They also document that AirPods Max 2 firmware updates require iOS 27 Beta 2 or macOS 27 Beta 2 at minimum—a dependency to account for if you're working on accessory integration.

In isolation, Pace and Expressivity customization is a focused feature. In the context of iOS 27's full Siri rebuild, it points toward an assistant designed to adapt to individual users rather than present a single default voice. Whether Apple delivers the full vision by release remains an open question, but the direction is clear.