[opus] Opus Tools -- low bitrates, new features in 1.5, "expect-loss"

[Petr Pařízek]

> I mean the original sweep _as_encoded_, sorry.

> You're right, I have noticed this as well. In that case, I would find it just super-duper-important to have high-quality anti-aliasing filters if the bottom part of the spectrum needs to be downsampled before being encoded by the "almost SILK".
> I don't know the detailed format of the Opus data but my assumption is that encoding the bottom part at the original sample rate would take more bits than encoding it downsampled; but I'm just hypothesizing here.
> I didn't notice this "bug" before because at high bitrates, everything is apparently send to CELT (which operates at 48 kHz). Or at least that's how I've understood it; I should probably read a larger part of RFC 6716 to confirm this.
>
> Petr
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.xiph.org/pipermail/opus/attachments/20240809/71d7c3c6/attachment.htm>