Question about the returns of CutSet.mix() ?

CutSet.mix() returns only MixedCut when p=1, but returns both MonoCut and MixedCut when p <1.0;

from lhotse import Recording, CutSet

 wav_file = "a.wav"
 noise_file = "b.wav"

 wav_cut = Recording.from_file(wav_file).to_cut()
 noise_cut = Recording.from_file(noise_file).to_cut()

 wav_cuts = CutSet.from_cuts([wav_cut])
 noise_cuts = CutSet.from_cuts([noise_cut])

 for p in [0.1, 1.0]:
     mixed_cuts = wav_cuts.mix(
         cuts=noise_cuts,
         duration=None,
         snr=10,
         mix_prob=p,
         preserve_id=None,
         seed=42,
         random_mix_offset=True,
     ).to_eager()

     print(f"Mixed cuts lengths: {len(mixed_cuts)}")

outputs

Mixed cuts lengths: 2
Mixed cuts lengths: 1

Hi, @pzelasko Is it a bug here ? As what i want is only the mixed cut MixedCut when training ASR model. And when training ASR model with the lhotse.dataset.CutMix transform, the time costs 2x compared to the training without lhotse.dataset.CutMix transform.

lhotse-speech / lhotse

Question about the returns of CutSet.mix() ? #1267