Is OPUS 1.5.2 Deep PLC applicable for decoding 48kHz stereo audio data?

Hi Team,

I recently tested OPUS 1.5.2 Deep PLC using opus_demo. Based on both MOS values (calculated with the ViSQOL tool) and subjective listening, I found that Deep PLC does not show noticeable improvement over legacy PLC in Opus 1.3 when decoding 48kHz stereo audio. For 16kHz mono audio, the MOS value is slightly higher with Deep PLC, but the difference is not perceptible by listening.

I have a few questions:

1. Is Deep PLC applicable for decoding 48kHz stereo audio data?
2. For 16kHz mono audio, should there be a significant improvement in audio quality by listening?
3. Are there any issues with my testing steps below?

My testing steps:

• 48kHz, stereo
  Encode:
  ./opus_demo -e audio 48000 2 64000 -framesize 10 speech_48k_stereo.pcm output.opus
  Decode:
  ./opus_demo -d 48000 2 -loss 20 -dec_complexity 10 output.opus output.pcm

• 16kHz, mono
  Encode:
  ./opus_demo -e voip 16000 1 24000 -cbr -framesize 20 -complexity 3 -loss 0 speech_16k_mono.pcm output.opus
  Decode:
  ./opus_demo -d 16000 1 -loss 20 -dec_complexity 10 output.opus output.pcm

Thank you for your help!

Start with the two opus_demo command sequences in the issue and review the Deep PLC and legacy PLC documentation or relevant codec tests. Reproduce the 48 kHz stereo and 16 kHz mono comparisons with ViSQOL and listening checks. Done means documenting whether Deep PLC applies, what improvement is expected, and whether the testing steps are valid.