# ai-coustics Docs

> Integrate and operate ai-coustics audio models in your application.

- [Overview](https://docs.ai-coustics.com/index.md): Integrate speech enhancement, voice activity detection and audio analysis into your application.
- [SDK quickstart](https://docs.ai-coustics.com/models/get-started/sdk-quickstart.md): Install the ai-coustics SDK, process audio and check the result in your preferred language.
- [LiveKit quickstart](https://docs.ai-coustics.com/models/get-started/livekit-quickstart.md): Add dedicated voice activity detection and Tyto analysis to a LiveKit room with complete Python and Node.js examples.
- [Pipecat quickstart](https://docs.ai-coustics.com/models/get-started/pipecat-quickstart.md): Run a local audio loopback with speech enhancement, verify activation and add standalone voice activity detection.
- [Quickstart audio fixture](https://docs.ai-coustics.com/guides/test-audio.md): Download a short, attributed speech recording for repeatable SDK setup checks.
- [Python](https://docs.ai-coustics.com/reference/sdk/language-bindings/python.md): Process a WAV with Python, then adapt the SDK to synchronous or async audio streams.
- [Node.js](https://docs.ai-coustics.com/reference/sdk/language-bindings/nodejs.md): Process a WAV with Node.js, then integrate synchronous or async native audio processing.
- [C](https://docs.ai-coustics.com/reference/sdk/language-bindings/c.md): Build a C WAV-processing program with explicit SDK error checks and native cleanup.
- [C++](https://docs.ai-coustics.com/reference/sdk/language-bindings/cpp.md): Build a C++ WAV-processing program with checked results and scoped native lifetimes.
- [Rust](https://docs.ai-coustics.com/reference/sdk/language-bindings/rust.md): Build a Rust WAV-processing program with the published crate and explicit native linking.
- [WebAssembly](https://docs.ai-coustics.com/reference/sdk/language-bindings/wasm.md): Process browser audio with the released WASM package and keep SDK keys on your backend.
- [Authenticate apps](https://docs.ai-coustics.com/models/get-started/authenticate-apps.md): Keep SDK keys on trusted servers and manage short-lived credentials for supported client integrations.
- [Pricing](https://docs.ai-coustics.com/models/get-started/pricing.md): Find SDK trial, usage, billing and support information.
- [Audio format](https://docs.ai-coustics.com/reference/concepts/audio-format.md): Prepare mono PCM audio for every audio entry point in the SDK.
- [Streams and state](https://docs.ai-coustics.com/reference/concepts/streams-and-state.md): How SDK objects hold per-stream state, when to reset it and which calls are real-time safe.
- [Latency](https://docs.ai-coustics.com/reference/concepts/latency.md): Separate SDK output delay from execution time, queueing and end-to-end application latency.
- [How to evaluate audio processing](https://docs.ai-coustics.com/guides/evaluate-audio.md): Compare paired recordings, downstream accuracy, agent behavior and runtime cost with a reproducible procedure.
- [Performance](https://docs.ai-coustics.com/reference/concepts/performance.md): Measure processing time, capacity and queue behavior on your deployment hardware.
- [How to deploy the SDK](https://docs.ai-coustics.com/production/deployment.md): Prepare model files, credentials, network access and runtime checks for a production deployment.
- [AirTen](https://docs.ai-coustics.com/reference/concepts/airten.md): The CPU inference runtime used by the ai-coustics SDK.
- [SDK telemetry](https://docs.ai-coustics.com/reference/concepts/sdk-telemetry.md): Separate local inference, authorization, usage reporting, error reporting and optional metrics export.
- [OpenTelemetry](https://docs.ai-coustics.com/models/get-started/opentelemetry.md): Export SDK operational metrics to your own observability backend via OpenTelemetry.
- [How to troubleshoot an integration](https://docs.ai-coustics.com/production/troubleshooting.md): Find the failing stage, apply a targeted recovery and verify that audio processing is active.
- [Models](https://docs.ai-coustics.com/reference/sdk/models.md): Choose a model by task and verify model IDs, native audio geometry and SDK compatibility.
- [Quail Voice Focus](https://docs.ai-coustics.com/models/voice-focus/quail-voice-focus.md): Primary-speaker isolation for speech-to-text and voice agents.
- [Quail Multi Speaker](https://docs.ai-coustics.com/models/speech-enhancement/quail-multi-speaker.md): Multi-speaker speech enhancement for speech-to-text and voice agents.
- [Voice Focus for voice AI systems](https://docs.ai-coustics.com/models/voice-focus/voice-focus-for-voice-ai-systems.md): Tune primary-speaker isolation against transcription errors and speech retention.
- [Speech enhancement for voice AI systems](https://docs.ai-coustics.com/models/speech-enhancement/speech-enhancement-for-voice-ai-systems.md): Evaluate Quail models with your speech-to-text provider and tune enhancement without losing target speech.
- [Tyto](https://docs.ai-coustics.com/models/audio-insight/tyto.md): Use audio risk and quality indicators to investigate downstream voice AI failures.
- [Real-time analysis with Tyto](https://docs.ai-coustics.com/models/audio-insight/real-time-analysis.md): Collect a live audio stream and run Tyto analysis outside the audio callback.
- [Batch call analysis with Tyto](https://docs.ai-coustics.com/models/audio-insight/batch-call-analysis.md): Analyze local recordings with the SDK and write scores for review.
- [VAD Multi Speaker](https://docs.ai-coustics.com/models/voice-activity-detection/vad-multi-speaker.md): Detect speech activity from any speaker in the original audio.
- [VAD Voice Focus](https://docs.ai-coustics.com/models/voice-activity-detection/vad-voice-focus.md): Detect speech activity from the primary speaker for voice agents.
- [How to use and tune VAD](https://docs.ai-coustics.com/models/voice-activity-detection/vad.md): Process original audio and tune speech boundaries with VAD Multi Speaker or VAD Voice Focus.
- [Rook Multi Speaker](https://docs.ai-coustics.com/models/perceptual-speech-enhancement/rook-multi-speaker.md): Speech enhancement for human listening in calls and other real-time audio applications.
- [Deprecated models](https://docs.ai-coustics.com/models/older-models/deprecated-models.md): Guide to migrating away from deprecated models.
- [Model naming changes](https://docs.ai-coustics.com/models/older-models/model-naming-changes.md): Update model names and IDs for the August and September 2026 transitions.
- [Language bindings](https://docs.ai-coustics.com/reference/sdk/language-bindings.md): Choose a released binding and follow a complete integration example.
- [Examples](https://docs.ai-coustics.com/reference/sdk/examples.md): Browse example code for the ai-coustics SDK across all supported languages.
- [Compatibility matrix](https://docs.ai-coustics.com/reference/sdk/compatibility-matrix.md): Compatibility matrix for core SDK, language bindings and model file versions.
- [Python API reference](https://docs.ai-coustics.com/reference/sdk/api/python/index.md): Complete public API surface for aic-sdk 3.3.0: models, enhancement, VAD, analysis and errors.
- [Python models and configuration](https://docs.ai-coustics.com/reference/sdk/api/python/models-and-config.md): Python reference for Model, ProcessorConfig, OtelConfig, get_sdk_version and get_compatible_model_version in aic-sdk 3.3.0.
- [Python enhancement](https://docs.ai-coustics.com/reference/sdk/api/python/enhancement.md): Python reference for Processor, ProcessorAsync, ProcessorContext and ProcessorParameter in aic-sdk 3.3.0.
- [Python voice activity detection](https://docs.ai-coustics.com/reference/sdk/api/python/vad.md): Python reference for Vad, VadAsync, VadContext, EnergyVadContext and VadParameter in aic-sdk 3.3.0.
- [Python audio analysis](https://docs.ai-coustics.com/reference/sdk/api/python/analysis.md): Python reference for analyzer_pair, Collector, Analyzer, FileAnalyzer and AnalysisResult in aic-sdk 3.3.0.
- [Python errors](https://docs.ai-coustics.com/reference/sdk/api/python/errors.md): Exception constructors, attributes, causes and recovery for aic-sdk 3.3.0.
- [Node.js API reference](https://docs.ai-coustics.com/reference/sdk/api/node/index.md): Complete reference for the released Node.js SDK, including methods, parameters, errors and ownership.
- [Node.js models and configuration](https://docs.ai-coustics.com/reference/sdk/api/node/models-and-config.md): Load model weights, inspect SDK versions and configure optional OpenTelemetry export.
- [Node.js enhancement API](https://docs.ai-coustics.com/reference/sdk/api/node/enhancement.md): Process mono audio synchronously or asynchronously and control enhancement state.
- [Node.js VAD API](https://docs.ai-coustics.com/reference/sdk/api/node/vad.md): Read dedicated speech predictions and configure voice activity detection.
- [Node.js analysis API](https://docs.ai-coustics.com/reference/sdk/api/node/analysis.md): Collect mono audio and analyze the retained window with Tyto.
- [Node.js errors](https://docs.ai-coustics.com/reference/sdk/api/node/errors.md): Identify thrown errors and rejected promises, then recover with the correct state and inputs.
- [C API reference](https://docs.ai-coustics.com/reference/sdk/api/c/index.md): Complete public C SDK 0.25.0 functions, handles, structures and enums.
- [C models and configuration](https://docs.ai-coustics.com/reference/sdk/api/c/models-and-config.md): Load model data, inspect native audio metadata and configure optional OpenTelemetry.
- [C enhancement and processor control](https://docs.ai-coustics.com/reference/sdk/api/c/processing.md): Create and initialize an enhancement processor, process mono audio and control stream state.
- [C voice activity detection](https://docs.ai-coustics.com/reference/sdk/api/c/vad.md): Process a dedicated VAD model and read its probability and filtered speech decision.
- [C audio analysis](https://docs.ai-coustics.com/reference/sdk/api/c/analysis.md): Collect audio features and run analysis on a separate thread.
- [C errors and types](https://docs.ai-coustics.com/reference/sdk/api/c/errors-and-types.md): Look up public error codes, pointer ownership and every analysis field.
- [C++ API reference](https://docs.ai-coustics.com/reference/sdk/api/cpp/index.md): Complete C++11 wrapper 0.25.0 types, members, factories and ownership contracts.
- [C++ models and configuration](https://docs.ai-coustics.com/reference/sdk/api/cpp/models-and-config.md): Load model data and inspect audio metadata and configuration values.
- [C++ enhancement and processor control](https://docs.ai-coustics.com/reference/sdk/api/cpp/processing.md): Create a dedicated enhancement object and manage its stream state.
- [C++ voice activity detection](https://docs.ai-coustics.com/reference/sdk/api/cpp/vad.md): Create a dedicated VAD object and manage its stream state.
- [C++ audio analysis](https://docs.ai-coustics.com/reference/sdk/api/cpp/analysis.md): Collect mono audio and compute a checked Tyto analysis result.
- [C++ results and errors](https://docs.ai-coustics.com/reference/sdk/api/cpp/results-and-errors.md): Check factory results, move owned values and interpret every SDK error.
- [Rust API reference](https://docs.ai-coustics.com/reference/sdk/api/rust/index.md): Complete application API for aic-sdk 0.25.0, with ownership, feature gates, errors and runtime linking.
- [Rust models and configuration](https://docs.ai-coustics.com/reference/sdk/api/rust/models-and-config.md): Model loading, borrowed data, configuration fields and SDK metadata in aic-sdk 0.25.0.
- [Rust enhancement](https://docs.ai-coustics.com/reference/sdk/api/rust/enhancement.md): Processor, ProcessorAsync, contexts, stream ownership and parameters in aic-sdk 0.25.0.
- [Rust voice activity detection](https://docs.ai-coustics.com/reference/sdk/api/rust/vad.md): Vad, VadAsync, EnergyVadContext, prediction control and parameters in aic-sdk 0.25.0.
- [Rust audio analysis](https://docs.ai-coustics.com/reference/sdk/api/rust/analysis.md): Collector, Analyzer, FileAnalyzer and AnalysisResult in aic-sdk 0.25.0.
- [Rust errors and features](https://docs.ai-coustics.com/reference/sdk/api/rust/errors-and-features.md): Cargo features, runtime loading, complete error variants and trait contracts for aic-sdk 0.25.0.
- [WebAssembly API reference](https://docs.ai-coustics.com/reference/sdk/api/wasm/index.md): Complete application API for the released WebAssembly package.
- [WebAssembly initialization and lifecycle](https://docs.ai-coustics.com/reference/sdk/api/wasm/initialization.md): Initialize the module, choose an input source and manage owned handles.
- [WebAssembly model API](https://docs.ai-coustics.com/reference/sdk/api/wasm/models-and-config.md): Load model bytes and inspect model metadata.
- [WebAssembly enhancement API](https://docs.ai-coustics.com/reference/sdk/api/wasm/enhancement.md): Process audio blocks and control enhancement state.
- [WebAssembly VAD API](https://docs.ai-coustics.com/reference/sdk/api/wasm/vad.md): Process dedicated VAD models and read speech predictions.
- [WebAssembly analysis API](https://docs.ai-coustics.com/reference/sdk/api/wasm/analysis.md): Collect audio, analyze a window and release analysis results.
- [WebAssembly errors](https://docs.ai-coustics.com/reference/sdk/api/wasm/errors.md): Handle initialization, model, parameter and processing failures.
- [SDK and model versions](https://docs.ai-coustics.com/reference/sdk/upgrading.md): How core SDK releases and model file versions fit together.
- [Which LiveKit plugin to use](https://docs.ai-coustics.com/reference/livekit/plugin-scope.md): Choose the official plugin for speech enhancement or the ai-coustics-maintained plugin for dedicated VAD and Tyto.
- [Migrate from livekit-plugins to livekit-extras](https://docs.ai-coustics.com/reference/livekit/migrate-to-livekit-extras.md): Move dedicated VAD and Audio Insight from the ai-coustics-maintained plugin 0.2.0 to ai-coustics LiveKit extras 0.25.0.
- [MCP server and text export](https://docs.ai-coustics.com/reference/agents/mcp-server.md): Connect a coding agent to canonical documentation and retrieve pages as Markdown.
- [Migrate to Mono Processing](https://docs.ai-coustics.com/reference/deprecated/multi-channel-to-mono.md): Update SDK integrations from the multi-channel buffer APIs to the mono processing and buffering API.
- [Migrate to the Dedicated VAD](https://docs.ai-coustics.com/reference/deprecated/energy-vad-to-dedicated-vad.md): Move voice activity detection off the processor and onto a dedicated VAD object backed by a VAD model.
- [Migrate Python SDK 2.5 to 3.0](https://docs.ai-coustics.com/reference/deprecated/python-2-5-to-3-0.md): Upgrade to ai-coustics Python SDK 3.0 with mono-only processing, a dedicated Vad class and renamed configuration fields and errors.
- [Changelog](https://docs.ai-coustics.com/changelog.md): Product updates and announcements


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.