需要一个使用 react-native-audio-api 库的 React Native 项目。不附带脚本，技能仅为说明文档。它要求代理通过 webfetch 获取参考文件中链接的文档页面，因此获取 API 签名和配置选项需要网络访问。
Audio
Software Mansion's production audio patterns for React Native using react-native-audio-api.
Load at most one reference file per question. For API signatures and config options, webfetch the documentation pages linked in each reference file.
Critical Rules
Single AudioContext: Create one AudioContext instance per app. Multiple instances lead to conflicting states (one running, another suspended). Encapsulate it in a singleton class.
Single AudioRecorder: Create one AudioRecorder instance and reuse it. Switching between recorder instances has noticeable impact on performance, memory, and battery.
Suspend when idle: A running AudioContext plays silence even with no source nodes connected, draining battery. On iOS, it also prevents the lock screen from showing a paused state. Use suspend() when audio is temporarily not needed, close() when done permanently.
AudioBufferSourceNode is single-use: An AudioBufferSourceNode can be started only once. To replay the same sound, create a new node. Nodes are inexpensive to create; reuse the AudioBuffer.
Use AudioParam scheduling for smooth changes: Direct, immediate changes to gain, frequency, or other params cause audible clicks. Use linearRampToValueAtTime, exponentialRampToValueAtTime, or setTargetAtTime for smooth transitions.
Mutate typed arrays in place for visualizations: When passing audio data to Reanimated shared values, use .modify() to mutate the existing Float32Array rather than allocating a new one each frame. Allocating at 60fps+ causes GC jank.
References
File
When to read
audio.md
Decision tree for choosing an audio approach; AudioContext lifecycle and singleton pattern; audio graph concepts; decoding audio data; AudioBuffer state management
playback.md
Playing audio with AudioBufferSourceNode, OscillatorNode, StreamerNode, AudioBufferQueueSourceNode; looping and scheduling; AudioParam scheduling methods; playback rate and pitch; noise generation
recording.md
Recording audio with AudioRecorder; three recording modes (file, data callback, graph processing); file output configuration and formats; permissions; background recording setup
effects-and-analysis.md
Audio effects chain (GainNode, BiquadFilterNode, DelayNode, ConvolverNode, StereoPannerNode, WaveShaperNode); ADSR envelopes; AnalyserNode for time-domain and frequency-domain data; audio visualization patterns
worklets.md
Audio worklets with WorkletNode, WorkletSourceNode, WorkletProcessingNode; UIRuntime vs AudioRuntime; performance budgets and latency constraints; custom synthesis and real-time processing
system-and-notifications.md
AudioManager for session configuration and system events; PlaybackNotificationManager and RecordingNotificationManager for media controls; permissions; interruption handling; testing with mocks
