I tried it myself and have not run into the issue you're describing. Neither with the audio files provided in the SFML examples, nor with other stereo and mono files, nor with different output formats OGG, WAV, FLAC.
This is the code I used:
#include <SFML/Audio.hpp>
int main()
{
sf::SoundBuffer buf;
buf.loadFromFile("input.wav");
std::vector<sf::Int16> samples(buf.getSampleCount());
for(std::size_t i = 0; i < buf.getSampleCount(); i++)
{
samples[i] = buf.getSamples()[i];
}
sf::SoundBuffer buff;
buff.loadFromSamples(samples.data(), buf.getSampleCount(), buf.getChannelCount(), buf.getSampleRate());
buff.saveToFile("output.wav");
}
I have however run into a memory allocation issue with the OGG writer, but this is unrelated to your problems.