Sample rate conversion is algorithm of changing number samples (measurements) of signal level per second without changing waveform of the signal.
This kind of signal processing quality loss always.
In resulting signal appear non-linear distortions (noise burned by signal).
In high qualitative sample rate conversion algorithm the noise has level low -160 dB.
As comparing, studio audio cards has noise level about -120 dB.
Therefore noises of audio card masks conversion noise.
The algorithm steps are:
1) Increase up of input sampling rate to frequency, multiple output signal's sampling frequency.
2) Filtration of "parasitic" signals (non-linear distortions named as "artefacts") above half of target sample frequency.
3) Decreasing sampling rate as multiple rejection of samples.
Sampling rate conversion algorithm details described here...