Roformer.cpp/include/bs_roformer/audio.h

#pragma once
#include <vector>
#include <string>
#include <stdexcept>

/**
 * Audio buffer structure for storing audio data.
 * Data is stored in interleaved format (L, R, L, R, ...) for stereo.
 */
struct AudioBuffer {
    std::vector<float> data;  // Interleaved samples
    unsigned int channels;
    unsigned int sampleRate;
    size_t samples;           // Total samples (frames * channels)
};

/**
 * Audio file I/O utilities.
 * Supports any common audio format (WAV, MP3, FLAC, OGG, etc.) via FFmpeg/libav.
 * Automatically resamples to target sample rate if needed.
 */
class AudioFile {
public:
    /**
     * Load audio from any common audio file format.
     * Audio is automatically resampled to 44100 Hz and converted to stereo float32.
     * @param path Path to the audio file
     * @param target_sample_rate Target sample rate (default: 44100 Hz)
     * @return AudioBuffer containing the loaded audio data
     * @throws std::runtime_error if the file cannot be opened or decoded
     */
    static AudioBuffer Load(const std::string& path, int target_sample_rate = 44100);
    
    /**
     * Save audio to a WAV file (PCM float32).
     * @param path Path to save the WAV file
     * @param buffer AudioBuffer containing audio data to save
     * @throws std::runtime_error if the file cannot be written
     */
    static void Save(const std::string& path, const AudioBuffer& buffer);
};