[flac-dev] [PATCH] stream_encoder : Improve selection of residual accumulator width

[lvqcl]

Miroslav Lichvar wrote:

> Adding the extra bits to bps in evaluate_fixed_subframe_ instead of
> precompute_partition_info_sums_ is ok with me, that's what I suggested
> in the original thread discussing this problem,

found it: http://lists.xiph.org/pipermail/flac-dev/2013-July/004303.html

> but I think it should
> be done also in the lpc function. Adding order instead of 4 might work
> for fixed frames, for LPC it looks too pessimistic.

I'm OK with any version.

> To me it looks like (order * subframe_bps) in the calculation is the
> number of bits for the warmup samples in the fixed subframe.

Thanks!