Question: DRED decoding

[jacksongoode]

I'm curious if you have tried testing DRED encoding/decoding successfully? I'm finding a working example of this hard to come by. If you have, it might be nice in an example in this repo.

[DoumanAsh]

Can you elaborate a bit more?

Here is outline about DRED:

• Idea is to have encoder add additional redundancy within bitstream's padding to enable decoder to re-construct bitstream in case data loss (you have no benefits of DRED if you do not expect data loss)
• DRED purpose is to ensure quality of voice data, not general purpose audio
• First you have to configure your encoder:
  • https://docs.rs/opusic-c/1.5.3/opusic_c/struct.Encoder.html#method.set_dred_duration
  • I believe you should not use opus's default error correction mechanism (i.e. https://docs.rs/opusic-c/1.5.3/opusic_c/struct.Encoder.html#method.set_inband_fec should be used with InbandFec::Off which is default value)
  • You most likely want to use VOIP mode when configuring encoder (as that's main target of DRED) https://docs.rs/opusic-c/1.5.3/opusic_c/enum.Application.html#variant.Voip
• As you notice from API, there is dedicated high level abstraction for DRED state and you must use one of its methods to decide data:
  • https://docs.rs/opusic-c/1.5.3/opusic_c/dred/struct.Dred.html#method.decode_to

I only made a very simple test case to verify dred works
https://github.com/DoumanAsh/opusic-c/blob/master/tests/opus.rs#L56-L65

But I actually do not use DRED myself since opus itself has builtin error correction mechanisms and DRED is pretty heavy code size wise
Still if you need some further advice let me know

[DoumanAsh]

Still as you mention it would be beneficial to add some example to dred module documentation
https://docs.rs/opusic-c/1.5.3/opusic_c/dred/index.html

As I do not use DRED myself I never actually bothered to document it properly and only added it for completeness sake...

[jacksongoode]

I see, the conditions required to enable it are quite unclear, thank you for adding some documentation here!

I suppose it might be helpful to add some encode/decode example with synthetic loss, I believe the strategy of backfilling the audio packets of a missing gap with a buffer is a little non-intuitive. I'd be happy to contribute a basic example if I can get it working.

[DoumanAsh]

The simplest approach would be to have voice clip, encode it with DRED configured, and then decode it by skipping some packets in the middle (this should only work with speech I believe)
But I'm not sure you can get deterministic result.

[DoumanAsh]

On side note you can find RFC draft for DRED here
https://datatracker.ietf.org/doc/draft-ietf-mlcodec-opus-dred/

DRED works by having the encoder transmit acoustic features in the
Opus bitstream. On the receiver side, if packets are lost, then the
first packet to arrive will contain the acoustic features for a
certain duration in the past. The decoder can then use the features
to synthesize the missing speech -- either from the last received or
from the last audio samples produced by packet loss concealment
(PLC). Although the synthesized speech samples should be consistent
with the last known samples at the point of the transition, the
features do not contain waveform-specific or phase-specific
information so the synthesized speech waveform will significantly
deviate from the original waveform, despite sounding similar.