Apple seeded iOS 27 developer beta 3 (build 24A5380h) on July 6, 2026, and its headlining addition is functional: the Siri voice customization controls for Pace and Expressivity, which had been visible but gated as "Coming soon" in earlier betas, are now fully operative.

Both controls surface as sliders. Pace governs how quickly or slowly Siri speaks; Expressivity governs the degree of affective coloring in the voice — how much human-like emotional variance the synthesized output carries. As a user drags either slider, Siri reads a sample phrase ("You have one new message") in real time so the adjustment is immediately audible rather than abstract. That preview loop is a small but deliberate UX choice: it removes the need to commit a setting, leave the menu, trigger Siri, and then return to tune again.

The features were announced at WWDC 26 in June 2026, per Apple's WWDC session on Siri and iPhone, alongside a broader repositioning of Siri as an AI-native assistant. iOS 27 extends Siri's activation surface considerably: in addition to the classic voice trigger and side-button press, users can now invoke Siri by swiping down from the Dynamic Island, typing, or launching a dedicated standalone Siri app. Beta testers also have access to a range of Siri voices across different accents, selectable within the same settings panel.

What the Controls Actually Mean

Pace and Expressivity are not novel concepts in speech synthesis — text-to-speech systems have exposed speed and prosody parameters for years, often via developer APIs. What Apple is doing here is surfacing those parameters directly in consumer UI, without requiring any third-party app or accessibility workaround. That matters for a meaningful segment of users: people with auditory processing differences, non-native speakers of the device's set language, and anyone who simply finds the default cadence uncomfortable over extended interaction.

Expressivity is the more technically interesting of the two. Prosody manipulation in neural TTS pipelines involves more than tempo scaling; it typically requires inference-time conditioning on affective embeddings. A slider that modulates that parameter in real time implies the voice model running on-device — or via a low-latency server path — is conditioned to accept that input continuously rather than baking a single output style at generation time. Apple has not published technical specifics on the implementation, but the live-preview behavior suggests the inference path is fast enough to be practically synchronous.

Beta Caveats Worth Noting

TechCrunch reported that some users on X described losing access to the new AI-powered Siri after updating to beta 3, with others seeing their device re-index data post-install. Both are consistent behaviors for a beta that is making substantive changes to on-device model storage and indexing pipelines.

Apple's own beta release notes carry a separate flag worth keeping in mind for privacy-conscious developers and users: a known issue warns that disabling Siri may not reliably delete stored Siri and Dictation interaction data. That is a bug, not a feature, but it is the kind of bug that warrants watching through subsequent betas given the regulatory attention on voice assistant data retention in multiple jurisdictions.

Beta 3 also ships an updated Reminders app icon — a minor cosmetic change, noted here only because icon refreshes in late betas sometimes signal a broader design sweep incoming before GM.

Where This Sits in the Broader iOS 27 Picture

The AI-powered Siri that iOS 27 introduces is architecturally different from its predecessor. Apple's developer page describes Siri's language understanding as something that "evolves" in this release — careful language that sidesteps specifics but points to a model that updates or adapts rather than remaining static between OS versions. The expanded activation methods and now the voice customization layer are all UI expressions of a back-end that Apple has been rebuilding since at least WWDC 25.

For developers, the practical near-term item is the release notes on developer.apple.com, which also document that AirPods Max 2 firmware beta updates require iOS 27 Beta 2 or macOS 27 Beta 2 as a minimum — a dependency to factor into any accessory integration work.

Pace and Expressivity customization joining the public-facing settings panel is, in isolation, a focused change. Set against the full iOS 27 Siri rebuild, it points toward a version of the assistant designed to be genuinely adapted to individual users rather than delivered as a single default persona. Whether Apple executes that fully by general availability is still an open question, but the direction is legible.