vocra_flutter 0.2.2

The voice layer for Flutter: talk to an AI that talks back — natural turn-taking, lifelike speech, streaming replies. On-device, your own keys. vocra.cloud

Changelog

0.2.2

Packaging and metadata only — no Dart API or behavior changes.

Added

• Swift Package Manager support on iOS: ios/vocra_flutter/Package.swift, with the native AEC sources moved to ios/vocra_flutter/Sources/vocra_flutter/. The CocoaPods podspec is still maintained, so apps that have not enabled SPM build unchanged.
• A funding link (GitHub Sponsors) and https://www.vocra.cloud in the package metadata.

Fixed

• repository and issue_tracker pointed at the demo app repository, which broke pub.dev's repository verification. Both now resolve to this package's directory in the SDK repository.

Changed

• Licensed under MIT (the 0.2.1 release was published as MIT; an interim proprietary relicense never shipped to pub.dev).
• Depends on vocra_core: ^0.2.2.

0.2.1

Updates the published 0.2.0 release with the accumulated app-facing API.

Added

• VocraSession (renamed from VoiceSession), built on VocraConfig.
• observe({onState, onMessages, onTranscript, onError, onMetrics}) — one-call stream wiring returning a cancelable VocraSubscription.
• messages — a live aggregated conversation stream (interims collapsed).
• conversation, endSession(), sessionEnded, lastReport — full conversation retrieval and a SessionReport on every end path.
• mute() / unmute() / isMuted and a surfaced interrupt().
• Surfaces the new vocra_core 0.2.1 features (provider facades, typed model/ voice catalogs, xAI + Z.ai LLMs, structured prompts, greeting, natural speech, session policies) through the app API.

Fixed

• Mic capture falls back to the hardware sample rate (48/44.1 kHz) with on-device downsampling to 16 kHz when the platform refuses direct 16 kHz capture — fixes "Format conversion is not possible" on the iOS simulator.
• A mic-resume failure after a turn now surfaces on errors instead of crashing.
• Ending a session mid-turn no longer leaves a restarted session's microphone permanently unable to reach speech recognition.

Changed

• Breaking: VoiceSession → VocraSession, VoiceConfig → VocraConfig.
• Depends on vocra_core: ^0.2.1.

Docs

• iOS setup now documents the required PERMISSION_MICROPHONE=1 Podfile macro — without it permission_handler never shows the mic prompt and start() fails.
• Branded README with the Vocra logo, badges, and a vocra.cloud website link.

0.2.0

Initial pub.dev release: VoiceSession app-facing API, mic capture (record), ordered playback (just_audio), microphone permissions, audio-session interruption/becoming-noisy handling, SecureKeyStore, an optional native echo-cancellation full-duplex module, and a runnable example app.