aic-sdk

What it is and what it does

aic-sdk is a Python wrapper around ai-coustics' compiled audio processing engine, providing access to real-time audio enhancement, voice activity detection (VAD), and analysis capabilities. It processes audio as 1D mono numpy arrays and requires a license key to initialize. The SDK supports both synchronous and asynchronous APIs, allowing single-block processing or concurrent batch operations. It depends only on numpy for array handling and includes pre-built wheels for Python 3.10–3.14 across macOS, Linux, and Windows architectures.

The package is designed for streaming audio workflows: you load a model from disk or download it from ai-coustics' CDN, configure a processor with sample rate and block size, and then feed audio blocks through it. The processor maintains internal state and can be controlled via a thread-safe context object. VAD and enhancement can run independently or in parallel on the same audio stream, and telemetry can be configured per-instance via OpenTelemetry settings.

Use it for:

• Real-time voice enhancement in VoIP or conferencing applications by processing audio blocks as they arrive.
• Voice activity detection in speech recognition pipelines to filter silence and reduce processing overhead.
• Audio quality analysis and diagnostics using the Tyto analysis model on recorded or streaming audio.
• Batch audio enhancement in post-processing workflows by downloading models and processing multiple files.
• Concurrent audio processing in multi-threaded applications using async APIs and processor contexts.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Python bindings for ai-coustics audio enhancement, voice activity detection, and analysis SDK, supporting real-time audio processing with numpy arrays.

Yes, if you need audio enhancement or voice activity detection and have a valid ai-coustics license. The SDK is actively maintained, permissively licensed, has no known vulnerabilities, and offers both sync and async APIs for flexible integration. Install friction is moderate due to compiled wheels, but platform coverage is broad. Requires Python >=3.10 and an environment variable for the license key.

Install

aic-sdk on PyPI

pip install aic-sdk

uv add aic-sdk

poetry add aic-sdk

Installing aic-sdk

Before you install

Medium install friction due to compiled wheels across multiple Python versions and platforms. Active maintenance with recent release (4 days old). Requires a license key from developers.ai-coustics.com to function.

License in practice

Licensed under Apache Software License (permissive), allowing commercial and private use with minimal restrictions.

Quickstart

pip install aic-sdk

import aic_sdk as aic
import numpy as np
import os

license_key = os.environ["AIC_SDK_LICENSE"]
model_path = aic.Model.download("quail-vf-2.2-l-16khz", "./models")
model = aic.Model.from_file(model_path)
config = aic.ProcessorConfig.optimal(model)
processor = aic.Processor(model, license_key, config)
audio_block = np.zeros(config.block_size, dtype=np.float32)
processed = processor.process(audio_block)

Requires Python >=3.10 and a valid ai-coustics license key set in the AIC_SDK_LICENSE environment variable.

Verify before relying

• Whether the SDK's telemetry (OpenTelemetry) is enabled by default and what data it collects.
• Performance characteristics and latency for real-time audio processing at different sample rates.
• Whether downloaded models are cached or re-downloaded on each call.
• Exact licensing terms and any restrictions on model redistribution or commercial deployment.

Package facts

Evidence: aic_sdk-3.1.0-cp310-cp310-macosx_10_12_x86_64.whl; aic_sdk-3.1.0-cp310-cp310-macosx_11_0_arm64.whl; aic_sdk-3.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; aic_sdk-3.1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; aic_sdk-3.1.0-cp310-cp310-win_amd64.whl; aic_sdk-3.1.0-cp310-cp310-win_arm64.whl; aic_sdk-3.1.0-cp311-cp311-macosx_10_12_x86_64.whl; aic_sdk-3.1.0-cp311-cp311-macosx_11_0_arm64.whl; aic_sdk-3.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; aic_sdk-3.1.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; aic_sdk-3.1.0-cp311-cp311-win_amd64.whl; aic_sdk-3.1.0-cp311-cp311-win_arm64.whl; aic_sdk-3.1.0-cp312-cp312-macosx_10_12_x86_64.whl; aic_sdk-3.1.0-cp312-cp312-macosx_11_0_arm64.whl; aic_sdk-3.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; aic_sdk-3.1.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; aic_sdk-3.1.0-cp312-cp312-win_amd64.whl; aic_sdk-3.1.0-cp312-cp312-win_arm64.whl; aic_sdk-3.1.0-cp313-cp313-macosx_10_12_x86_64.whl; aic_sdk-3.1.0-cp313-cp313-macosx_11_0_arm64.whl

Tags

More Speech packages