avioflow 0.1.1

High-performance audio decoding with FFmpeg and C++

AvioFlow

AvioFlow is a high-performance audio decoding and capture library powered by FFmpeg. It provides a modern C++20 API and seamless Python bindings for efficient audio processing, including file decoding, URL streaming, and low-latency system audio capture (WASAPI/DirectShow).

Features

• Multi-format Support: Decodes any format supported by FFmpeg (MP3, WAV, AAC, FLAC, etc.).
• Flexible Input: Load audio from files, memory buffers, network URLs, or custom streams.
• Hardware Capture:
  • WASAPI Loopback: Capture system output (what you hear).
  • DirectShow: Capture from microphones and other input devices.
• Resampling: Built-in support for target sample rate and channel conversion.
• Python Bindings: High-performance Python module using pybind11.

---

🛠 Build Instructions

Prerequisites

• Windows: Visual Studio 2022+, CMake 3.20+.
• Linux: GCC 11+, CMake 3.20+, FFmpeg development headers.

Build and Run

Use the provided PowerShell script for automated setup on Windows:

.\build.ps1

Or manually with CMake:

cmake -B build -DENABLE_PYTHON=ON -DENABLE_WASAPI=ON
cmake --build build --config Release

---

🚀 C++ Usage

Offline File Decoding

#include "avioflow-cxx-api.h"
#include <iostream>

int main() {
    avioflow::AudioStreamOptions options;
    options.output_sample_rate = 16000; // Resample to 16kHz
    options.output_num_channels = 1;    // Convert to Mono

    avioflow::AudioDecoder decoder(options);
    decoder.open("audio.mp3");

    auto meta = decoder.get_metadata();
    std::cout << "Codec: " << meta.codec << " Duration: " << meta.duration << "s" << std::endl;

    // Decode all at once
    auto samples = decoder.get_all_samples();
    std::cout << "Decoded " << samples.data[0].size() << " samples." << std::endl;

    return 0;
}

System Audio Capture (WASAPI)

decoder.open("wasapi_loopback");
while (!decoder.is_finished()) {
    auto frame = decoder.decode_next();
    if (!frame.data.empty()) {
        // Process real-time float samples...
    }
}

---

🐍 Python Usage

The Python package is built into build/bin/Release/avioflow. Add this path to your PYTHONPATH or copy the folder to your project.

Basic Decoding

import avioflow as av

# Set log level for debugging
av.set_log_level("warning")

# Initialize and open
decoder = av.AudioDecoder()
decoder.open("music.wav")

# Get info
meta = decoder.get_metadata()
print(f"Format: {meta.container}, {meta.sample_rate}Hz")

# Get all samples as nested lists [[ch1...], [ch2...]]
samples = decoder.get_all_samples()
print(f"Total samples: {len(samples.data[0])}")

Real-time Capture

# List available devices
devices = av.DeviceManager.list_audio_devices()
for d in devices:
    print(f"ID: {d.name}, Desc: {d.description}")

decoder = av.AudioDecoder()
decoder.open("audio=@device_cm_...") # Open microphone via DirectShow string

while True:
    frame = decoder.decode_next()
    if frame:
        # data is planar float32
        process(frame.data)

Installation for Developers

To use the local build in your script:

import sys
sys.path.append("path/to/avioflow/build/bin/Release")
import avioflow

---

License

MIT License