[FFmpeg-devel] [RFC] AAC encoder optimizations

Robert Swain robert.swain
Sun Sep 14 12:47:18 CEST 2008


2008/9/14 Kostya <kostya.shishkov at gmail.com>:
> On Sat, Sep 13, 2008 at 10:47:26PM -0700, Mike Melanson wrote:
>> Kostya wrote:
>>
>> > Since I can work on it mostly on spare time during work I would be grateful
>> > if you suggest some tests for quality. Oh, and headphones are one of the
>> > cheapest there.
>>
>> I know some audiophiles whose obsession with the perfect audio
>> experience borders on "mental disorder". But they have good ears and
>> better audio equipment. I wonder if it would be useful to enlist them to
>> compare audio output?
>
> Well, when I have encoding quality satisfiable for me at my home computer
> it will be the moment when testing can start.

I still recommend that we enlist users from the hydrogenaudio forums
to test for us. It's in their interest to have something competitive
with Nero/Quicktime.

> When encoder hits SVN we'll have plenty of testers too ;)
> And AAC is not for audiophiles anyway.

Not strictly true. Just because a codec is lossy does not mean it
cannot achieve audiophile quality - that is, transparency... sounding
the same as the original. When you've been saying optimal, I assume
you're referring to RD optimal. With an 'optimal' AAC encoder,
transparency should be easily achievable. I don't mean writing an
optimal encoder is easy.

I assume when tuning lossy audio codecs, people work on redistributing
bits to the tools/bitstream features that affect audible quality most
significantly so as to reduce the bitrate at which transparency is
achieved and at any given quality level and increase the audible
quality at a given bitrate.

Rob




More information about the ffmpeg-devel mailing list