Rockbox mail archiveSubject: Re: Need opinion about voicing for software codec targets
Re: Need opinion about voicing for software codec targets
From: Brandon Low <lostlogic_at_lostlogicx.com>
Date: Wed, 19 Apr 2006 21:41:04 -0500
On Thu, 04/20/06 at 01:34:10 +0200, Jens Arnold wrote:
> On 19.04.2006, Brandon Low wrote:
> > As for voice while paused, I do not promise this for a 3.0
> > feature, as it will involve one of many Bad Things. The
> > penalty choices for adding voice while paused basically amount
> > to: ~512k of RAM _and_ one of: between 0 and 1/4 second of
> What would the 512K RAM be used for?
To buffer voice into for playback, 512k is more than needed, sorry --
more like 150k to get almost 1 second of buffer space for voice to
> > lost audio, between 0 and 1/4 second of re-played audio or
> > some very complex partial pcm buffer DMA chunk logic. None of
> There's another option: Always pause at PCM buffer boundaries.
> The worst thing that might happen is that up to 1/4 second of
> voice may get lost/ re-played when pausing / unpausing while the
> voice is babbling.
You're right, there is the option of delaying pause by up-to 1/4 second.
> Regards, Jens
Received on 2006-04-20