[oliverlaz]

Ticket: https://linear.app/stream/issue/RN-468

🎯 Goal

Port of #3895 (the develop counterpart) to V10: the VoiceWave and VoiceGlow voice-agent visualizations in stream-chat-react-native/ai-components (and stream-chat-expo/ai-components), the useDictationAudioLevel hook, and the voice mode in ExpoChatGPTSample. Android, iOS and web already ship these visualizations in their AI components.

🛠 Implementation details

Everything under package/ is identical to #3895 (both of its commits, 0f07f0366 and 38a224a65, squashed into one), and applied to V10 without changes. See #3895 for the full description. In short:

• VoiceWave and VoiceGlow: one motion driver ported from Android, run as Reanimated worklets on the UI thread; the glow is Android's AGSL shader ported to Skia SkSL, the wave a Skia path. Props: state, colors, expressivity, audioLevel (SharedValue<number | null>) and stateLabels, plus View props.
• useDictationAudioLevel(), backed by a new AIDictationOnLevel event in package/shared-native/{android,ios}; useDictationState and useDictationTranscript are now exported.
• An optional voiceVisualization theme key (colors, backdropColor).
• A source option on useDictation, so ComposerView only fills its input from its own mic button.
• Fix: iOS dictation restores the AVAudioSession it found when dictation ends.

What differs from #3895 on V10 (all in examples/ExpoChatGPTSample):

• screens/ChatContent.tsx: V10 already moved channel creation and startAI/summarize from preSendMessageRequest into the client's sendMessageRequest handler in app/_layout.tsx, so the port only adds VoiceMode and useSendAIMessage there and drops the imports it no longer needs.
• components/VoiceMode/useVoiceConversation.ts: stream-chat v10 has no channel.state.latestMessages, so the voice mode reads channel.messagePaginator.items; and custom data lives under custom, so it checks message.custom?.ai_generated.
• hooks/useSendAIMessage.ts: the doc comment names the sendMessageRequest handler instead of preSendMessageRequest.

Docs: the v10-rc AI integration pages are updated in GetStream/getstream.io#810, together with v9.

🎨 UI Changes

Same UI as #3895; see its recordings.

🧪 Testing

• yarn lint (including validate-entry-isolation), cd package && yarn test:typecheck, yarn typecheck, yarn build and yarn test:unit (246 suites, 2302 tests) pass on this branch. shared-native:sync was run.
• The unit tests are the same as on feat(ai-components): add VoiceGlow and VoiceWave voice-agent visualizations #3895 (driver, bounce, palette, geometry, rendering, accessibility, theme fallbacks, reduced motion, dictation level).
• Not yet tested on a device on V10: the voice mode's switch to messagePaginator.items is type-checked only. Dictation needs a physical device (iOS 27 simulators can't initialize the speech recognizer).

☑️ Checklist

• I have signed the Stream CLA (required)
• PR targets the develop branch (targets V10; the develop counterpart is feat(ai-components): add VoiceGlow and VoiceWave voice-agent visualizations #3895)
• Documentation is updated (AI components README; v10-rc docs in GetStream/getstream.io#810)
• New code is tested in main example apps, including all possible scenarios
  • SampleApp iOS and Android
  • Expo iOS and Android

[Stream-SDK-Bot]

SDK Size

title	develop	branch	diff	status
js_bundle_size	2036 KB	2018 KB	-18022 B	🚀
ai_components_size	130 KB	163 KB	+33387 B	🔴

[stream-ci-bot]

🎉 This PR is included in version 10.0.0 🎉

The release is available on:

• v10.0.0
• GitHub release

Your semantic-release bot 📦🚀