Re: [codec] RTP Payload Format and File Storage Format for Opus Speech and Audio Codec - draft-spittka-payload-rtp-opus-00

["Benjamin M. Schwartz" <[e-mail]>]

On 07/05/2011 05:15 PM, Roni Even wrote:
> In general the payload specification should provide enough information that
> will enable the reader to create or parse the RTP payload without
> understanding the codec payload based on the payload specification. As such
> it is useful to provide information that will explain the general concepts
> of the codec and the optional parameters. So I find sections 1 and 3 useful
> and they should be in the document

Despite having removed Section 3 in my suggested edit, I agree.  The
payload document should present relevant information about Opus, such as
the available frame sizes and the significance of the low-bitrate
redundancy feature.  However, it should not discuss non-normative
behaviors of the reference coder, or internal details such as subsampling
and coding modes.

The most important thing to understand about Opus coding modes is that the
encoder can, and does, dynamically switch among them on a packet-by-packet
basis in order to optimize quality.  The draft seems to have been written
under the assumption that a stream, once started in a particular mode, may
not be able to switch to other modes.   This is no longer true, as
explained in the latest Opus draft.

--Ben