Arctan Voice Isolation

Arctan Voice Isolation is a Python SDK for real-time voice isolation in voice agents and audio pipelines.

It supports standard NumPy audio processing and integrations for LiveKit and Pipecat.

Installation

Install the base SDK:

pip install arctan-vi

Install a framework integration:

pip install "arctan-vi[livekit]"
pip install "arctan-vi[pipecat]"

Arctan Voice Isolation 0.8.1 supports CPython 3.10-3.14 on Linux x86-64 (glibc 2.28 or newer) and Windows x86-64. The Pipecat integration requires Python 3.11 or newer.

SDK key

Set your Arctan SDK key in the runtime environment:

export ARCTAN_SDK_KEY="arc_sk_live_..."

You can instead pass license_key="arc_sk_live_..." to a model, processor, or framework integration. An explicit license_key takes precedence over ARCTAN_SDK_KEY.

Standard Python

Process one stream

import numpy as np
import arctan

config = arctan.ProcessorConfig(sample_rate=24_000, num_channels=1)
processor = arctan.Processor(config=config)

try:
    audio = np.zeros((config.num_channels, config.num_frames), dtype=np.float32)
    enhanced = processor.process(audio)
finally:
    processor.close()

Processor.process() accepts finite float32 samples in [-1.0, 1.0], shaped as channels x frames. It returns a new float32 array with the same shape. Mono and stereo input are supported. Stereo is returned as dual-mono stereo.

The sample rate must remain stable for the lifetime of a processor. If num_frames is omitted, the SDK selects the native frame size. An explicit num_frames must contain a whole number of native processing blocks.

Process concurrent streams

For concurrent streams in one worker, share one Model and create one ProcessorAsync for each independent stream:

import asyncio
import numpy as np
import arctan


async def main() -> None:
    model = arctan.Model()
    configs = []
    processors = []

    try:
        for _ in range(2):
            config = arctan.ProcessorConfig.optimal(model)
            configs.append(config)
            processors.append(arctan.ProcessorAsync(model=model, config=config))

        frames = [
            np.zeros((config.num_channels, config.num_frames), dtype=np.float32)
            for config in configs
        ]
        enhanced = await asyncio.gather(
            *(
                processor.process_async(frame)
                for processor, frame in zip(processors, frames)
            )
        )
        print(len(enhanced))
    finally:
        for processor in processors:
            processor.close()
        model.close()


asyncio.run(main())

Use exactly one processor per call or audio stream. Keep frames ordered within each processor. Different processors may execute concurrently, but calls on a single ProcessorAsync are serialized. Close every processor before closing its shared model.

LiveKit

Install the LiveKit integration:

pip install "arctan-vi[livekit]"

Pass an Arctan noise canceller to LiveKit audio input options:

from arctan.livekit import arctan_enhancer
from livekit.agents import room_io

await session.start(
    agent=Assistant(),
    room=ctx.room,
    room_options=room_io.RoomOptions(
        audio_input=room_io.AudioInputOptions(
            noise_cancellation=arctan_enhancer.noise_canceller(),
        ),
    ),
)

Pass license_key= to noise_canceller() when the key is not provided through ARCTAN_SDK_KEY. Create a separate noise-canceller instance for every participant, call, or independent audio input.

Pipecat

Install the Pipecat integration:

pip install "arctan-vi[pipecat]"

Pass an Arctan audio filter to the transport:

from arctan.pipecat import arctan_enhancer
from pipecat.transports.base_transport import TransportParams

audio_filter = arctan_enhancer.ArctanAudioFilter()

transport_params = TransportParams(
    audio_in_enabled=True,
    audio_in_filter=audio_filter,
)

The Pipecat integration accepts mono PCM16 input and buffers arbitrary byte chunks until a complete processing block is available. Create a separate ArctanAudioFilter for every call or pipeline.