Re: [codec] Naming the modes

I like Voice and Universal as the API mode names.

For the internal mode names, Speech/Audio/Hybrid could be confusing because they're similar to the API modes, and because Hybrid is really also a Speech mode.
So maybe better with
- LPC/MDCT/Hybrid
- Predictive/Transform/Hybrid?

About having 40 and 60 ms modes for all sampling rates in Voice mode: we allow this already by combining multiple frames per packet (even though the current implementation doesn't do it yet).  Such multi-frame packets come with a few bytes overhead, but the assumption is that packets longer than 20 ms will be used mainly for lower bitrates where it's better to use the Predictive than the Hybrid mode.

koen.

----- Original Message -----
From: "Jean-Marc Valin" <jean-marc.valin@octasic.com>
To: bens@alum.mit.edu
Cc: codec@ietf.org
Sent: Thursday, December 2, 2010 7:45:13 AM
Subject: Re: [codec] Naming the modes

Note that these modes names are meant to describe now the codec works,
not necessarily the user-visible API.

Jean-Marc

On 10-12-02 10:25 AM, Benjamin M. Schwartz wrote:
> On 12/02/2010 06:59 AM, Jean-Marc Valin wrote:
>> Any thoughts?
>
> I think the codec has _two_ modes, not three: Voice and Universal.
> Voice mode operates at 8, 12, 16, 32, or 48 KHz. Universal mode
> operates at 8,
> 16, 32, or 48 KHz. The available window sizes depend on the mode and
> sample rate... although perhaps it would be better to abstract this
> away and allow 40 and 60 ms frames at all samplerates in Voice mode.
>
> The exact manner in which these sample rates are implemented is a
> detail irrelevant to the user.
>
> --Ben
>
>
>
>
> _______________________________________________ codec mailing list
> codec@ietf.org
> https://www.ietf.org/mailman/listinfo/codec

_______________________________________________ codec mailing list
codec@ietf.org
https://www.ietf.org/mailman/listinfo/codec