[CELT-dev] CELT in Digital Radio Mondiale

[Jean-Marc Valin]

Hi Michael,

Thanks for the link to your DRM work. I hope CELT works well for your 
application.

On 10-05-31 07:38 AM, Feilen, Michael wrote:
> Although the codec delivers very good audio quality and runs stable
> if the bitstream is error-free, the decoder (v0.7.1) produces a
> segfault under Windows in case the bitstream contains bit-errors.

I've done a quick investigation (need to do more), but it seems like the 
problem is related to frames of >512 samples, which are relatively new 
in CELT. If the problem is what I think, it occurs when one of the first 
four bits are corrupted. It shouldn't be too hard to fix (will do that 
ASAP).

> Currently, we detect faulty frames by an additional CRC check and
> avoid decoding them. I wonder if this behavior is intentional or if
> it might be a bug in the codec?

Once I fix this, what could actually be a good idea is to use some error 
correction codes, but only on the first 32-64 bits. These are the most 
important bits in CELT. We've done some experiments with SECDED codes in 
the past and using that can help bit error robustness quite a bit.

Cheers,

	Jean-Marc